[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-anthropic-aia-generation":3,"mdc--3tc3l6-key":37,"related-repo-anthropic-aia-generation":3773,"related-org-anthropic-aia-generation":3862},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":26,"repoUrl":27,"updatedAt":28,"license":29,"forks":30,"topics":31,"repo":32,"sourceUrl":35,"mdContent":36},"aia-generation","run AI impact assessments","Run an AI impact assessment — structured intake, risk analysis, regulatory classification per regime in scope, policy consistency diff, and recommendation with conditions. Uses the house-style structure learned from the seed impact assessment in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`. Use when user says \"impact assessment for\", \"assess this AI use case\", \"run an AIA\", \"generate an AIA\", \"we need to document this AI system\", \"AI risk assessment for X\", or follows a conditional triage result.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"anthropic","Anthropic","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fanthropic.png","anthropics",[13,17,20,23],{"name":14,"slug":15,"type":16},"Regulatory Compliance","regulatory-compliance","tag",{"name":18,"slug":19,"type":16},"Policy","policy",{"name":21,"slug":22,"type":16},"Legal","legal",{"name":24,"slug":25,"type":16},"Risk Assessment","risk-assessment",8721,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-for-legal","2026-05-13T06:03:19.61029",null,1642,[],{"repoUrl":27,"stars":26,"forks":30,"topics":33,"description":34},[],"A suite of plugins for legal workflows","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-for-legal\u002Ftree\u002FHEAD\u002Fai-governance-legal\u002Fskills\u002Faia-generation","---\nname: aia-generation\ndescription: >\n  Run an AI impact assessment — structured intake, risk analysis, regulatory\n  classification per regime in scope, policy consistency diff, and recommendation\n  with conditions. Uses the house-style structure learned from the seed impact\n  assessment in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`.\n  Use when user says \"impact assessment for\", \"assess this AI use case\", \"run an\n  AIA\", \"generate an AIA\", \"we need to document this AI system\", \"AI risk\n  assessment for X\", or follows a conditional triage result.\nargument-hint: \"[describe the use case or system, or pass a triage result]\"\n---\n\n# \u002Faia-generation\n\n1. Read `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`. Confirm impact assessment house style is populated.\n2. Determine risk track (fast or full) from governance tier and use case characteristics, using the framework below.\n3. Run intake — conversational, not a form.\n4. Regulatory classification for each regime in the footprint — research tier, prohibited-practice exposure, and applicable obligations; cite primary sources.\n5. Write assessment in house style (from seed doc, or default if none captured).\n6. Policy diff against `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` AI policy commitments.\n7. Output: assessment doc + conditions list + handoff flags (privacy PIA, vendor review if needed).\n\n```\n\u002Fai-governance-legal:aia-generation \"AI résumé screening for HR\"\n```\n\n---\n\n## Matter context\n\n**Matter context.** Check `## Matter workspaces` in the practice-level CLAUDE.md. If `Enabled` is `✗` (the default for in-house users), skip the rest of this paragraph — skills use practice-level context and the matter machinery is invisible. If enabled and there is no active matter, ask: \"Which matter is this for? Run `\u002Fai-governance-legal:matter-workspace switch \u003Cslug>` or say `practice-level`.\" Load the active matter's `matter.md` for matter-specific context and overrides. Write outputs to the matter folder at `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002Fmatters\u002F\u003Cmatter-slug>\u002F`. Never read another matter's files unless `Cross-matter context` is `on`.\n\n---\n\n## Purpose\n\nAn AI impact assessment is a documented decision, not a form. It answers: what\ndoes this AI system do, how does it reach its outputs, who's affected if it's\nwrong, what's the oversight, and is it okay to deploy. This skill structures that\nconversation and writes the output in this team's format — the one learned from the\nseed impact assessment during cold-start.\n\nAn AI impact assessment is not the same as a PIA. A PIA asks whether personal data\nis handled lawfully. An AIA asks whether the AI system is designed and deployed\nresponsibly. They often need to happen in parallel; they're not substitutes.\n\n## Load house style\n\nRead `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` → `## Impact assessment house style`. That has:\n- What triggers an impact assessment at this company\n- The structure template extracted from the seed assessment\n- Typical depth\n- Who signs off\n\nIf the seed structure is in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`, **use it**. The point is that this assessment\nlooks like the other assessments this team produces.\n\n**Jurisdictional scope.** This assessment applies the regulatory regimes listed in `## Regulatory footprint` in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`. AI legal rules, risk classifications, and deployment obligations vary materially by jurisdiction and are moving fast. If this system is (or will be) deployed outside that footprint, or if a choice-of-law question is in play, this analysis may not apply as written — re-run or expand the footprint.\n\n---\n\n## Step 0: Is an impact assessment needed?\n\nCheck the trigger criteria in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`.\n\n**Also check these regardless:**\n- Does this AI make or materially influence a decision affecting a person (employment,\n  credit, access, pricing, content moderation)?\n- Does this AI process personal data about individuals?\n- Is this a customer-facing AI system rather than purely internal?\n- Does this AI use a third-party model where the company is the deployer?\n- Is the use case in the elevated or high governance tier per `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`?\n\nIf none of the above and the house trigger isn't met:\n> \"Doesn't look like this needs a full impact assessment. Here's a one-paragraph\n> record for the file explaining why — in case anyone asks later.\"\n\n---\n\n## Step 1: Risk track\n\nBefore intake, determine which track to run. The tier definitions and the fast-track criteria come from `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` (`## Use case registry` and `## Governance tiers`), not from any hardcoded regime-specific framework.\n\nResearch the applicable risk classification framework for each regime in the user's regulatory footprint. Many regimes distinguish by risk tier, affected population, and decision consequentiality — research the specific criteria. Note that most regimes treat employee data as personal data and employee monitoring as consequential; don't assume internal-only systems are out of scope.\n\n> **No silent supplement.** If a research query to the configured legal research tool (Westlaw, EUR-Lex, regulator sites, or firm platform) returns few or no results for a regime's risk tiers or triggers, report what was found and stop. Do NOT fill the gap from web search or model knowledge without asking. Say: \"The search returned [N] results from [tool]. Coverage appears thin for [regime \u002F topic]. Options: (1) broaden the search query, (2) try a different research tool, (3) search the web — results will be tagged `[web search — verify]` and should be checked against the issuing authority before relying, or (4) flag as unverified and stop. Which would you like?\" A lawyer decides whether to accept lower-confidence sources.\n>\n> **Source attribution tiering.** Tag every citation in the AIA — regulatory text, delegated acts, guidance, standards — with its source. For model-knowledge citations, use one of three tiers rather than a single blanket \"verify\" tag:\n>\n> - `[settled]` — stable, well-known statutory and regulatory references unlikely to have changed (e.g., GDPR Art. 22 as a concept, the existence of Regulation (EU) 2024\u002F1689 as the EU AI Act). Still verify before certifying, but lower priority.\n> - `[verify]` — model-knowledge citations that are real but should be verified: specific delegated \u002F implementing acts, regulator guidance, NYC DCWP rules, Colorado AI Act provisions, harmonized standards, effective dates, EEOC guidance, and anything post-2023.\n> - `[verify-pinpoint]` — pinpoint citations (specific EU AI Act article numbers, annex references, Colorado AI Act subsections, NYC LL 144 rule sections, sub-paragraph letters) carry the highest fabrication risk and should ALWAYS be verified against a primary source. EU AI Act article numbers in particular shifted during consolidation; every pinpoint cite to the Act should be verified against the Official Journal text.\n>\n> Tool-retrieved citations keep their source tag (`[Westlaw]`, `[EUR-Lex]`, `[regulator site]`, or the MCP tool name); web-search citations remain `[web search — verify]`; user-supplied citations remain `[user provided]`. The tiering surfaces the real verification work — a reader who verifies everything verifies nothing. Never strip or collapse the tags.\n>\n> **For non-lawyer users, uncertain dates go in a confirm-list, not inline.** A `[verify]` tag on \"effective February 1, 2026\" reads as \"effective February 1, 2026\" to a CISO who doesn't know what `[verify]` means. Read `## Who's using this` in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`. If Role is **Non-lawyer** and a date, deadline, phase-in, threshold, or effective-date assertion is uncertain (would carry `[verify]` or `[verify-pinpoint]` if inline), replace the inline assertion with \"effective date: confirm with counsel\" (or \"threshold: confirm with counsel\", etc.) and collect all uncertain assertions in a final AIA section titled:\n>\n> > **Things I'm not certain about — ask your attorney to confirm before relying on this:**\n>\n> List each uncertain item there with (1) what I said, (2) what I'm uncertain about, (3) why it matters to the assessment. This prevents a non-lawyer reader from mistaking a flagged best-guess for a checked fact. Lawyer-role users get the inline `[verify]` treatment — they know what the tag means.\n\n**Fast track vs. full assessment:** `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` defines what qualifies for abbreviated treatment. If `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` doesn't define fast-track criteria, default to full assessment and ask the user what criteria they want captured for next time.\n\nIf in doubt, run the full assessment. A fast track that turns out to be wrong\nis worse than a thorough assessment on something low-risk.\n\n---\n\n## Step 2: Intake\n\nBefore writing anything, get answers to these. Conversational is fine — this\nis not a form to send them.\n\n### The system\n\n- What does the AI do? Describe it in plain language, not marketing copy.\n- Which model or vendor is powering it? Fine-tuned or off-the-shelf?\n- Where does it sit in the workflow — is it assistive (human reviews output),\n  augmentative (human can override but usually doesn't), or automated (no human\n  in the loop)?\n- What's the output — generated text, a score, a classification, a recommendation,\n  an action?\n\n### Who's affected\n\n- Who does the AI's output act on — employees, customers, third parties?\n- If the AI produces an error (false positive, false negative, hallucination), who\n  bears the harm and what's the worst realistic case?\n- Are any vulnerable groups disproportionately in scope — minors, job applicants,\n  people in financial distress, patients?\n\n### Inputs and data\n\n- What data does the AI take in?\n- Does it take in personal data? Whose?\n- Was the model trained on data from this company, or is it a foundation model\n  with no company-specific training?\n- Where does input data go — does it leave the perimeter to a third-party model\n  API?\n\n### Decisions and oversight\n\n- Does the AI output trigger an action automatically, or does a human decide what\n  to do with the output?\n- If there's human review: how often does the human actually change the AI's output?\n  (If the answer is \"rarely\" — the human isn't really reviewing; they're rubber-stamping.)\n- Is there an appeals or correction process for people affected by the AI's outputs?\n- Who is accountable for the AI system's outputs — is there a named owner?\n\n### Accuracy and failure\n\n- What's the known or estimated error rate? What testing has been done?\n- What happens when the AI is wrong — is the error surfaced, logged, corrected?\n- Has bias testing been done? Against what demographic groups?\n\n### Deployment stage and scale\n\nAsk:\n- **Stage:** \"Is this system (a) proposed and not yet built, (b) in pilot, (c) live in production, or (d) live and scaled?\"\n- **Scale:** \"Roughly how many individuals are affected per [month\u002Fyear]? How long has it been running?\"\n- **History:** \"Has it been assessed before? Has it produced decisions that were challenged, appealed, or reversed?\"\n\nStage changes the assessment: a proposed system gets a design review (can we build it safely?). A pilot gets a design review plus a \"before you scale\" gate. A live system gets a retrospective impact check (has it caused harm?) AND a go-forward review. A live-and-scaled system gets all of the above plus a remediation plan if issues are found, because you can't just turn it off.\n\n---\n\n## Step 3: Regulatory classification\n\n**Step 3 pre-check — footprint freshness.** Before iterating over the captured `## Regulatory footprint`, compare the use case's affected population and decision type (from Step 2) against the footprint as written. The footprint was set at cold-start, based on the company's operating posture at that moment. If the use case introduces an affected population (e.g., children, employees in a new state, EU data subjects) or a decision type (e.g., hiring, creditworthiness, health diagnosis, law enforcement, critical infrastructure) that the footprint does not contemplate, **re-derive the applicable regimes rather than iterating over the stale list.**\n\nSay to the user:\n\n> \"The practice profile's regulatory footprint was set for [affected populations \u002F decision types captured at cold-start]. This use case affects **[new population or decision type — e.g., employees in Colorado, minors under 13, credit decisions, biometric identification]**, which is not in the captured footprint. I'm going to re-derive the applicable regimes from the company's operating jurisdictions ([list from `## Company profile`]) and this use case's decision type ([Y]), rather than use the stale footprint. If this use case is representative of work you expect to see more of, update `## Regulatory footprint` at the end of this run so the next AIA doesn't have to re-derive.\"\n\nA common failure mode: the footprint lists EU AI Act + GDPR + NYC Local Law 144, and the use case is a hiring system being deployed into Illinois and Colorado. The footprint has no Illinois or Colorado entry, so iterating over it silently misses IL AIVIA, the new Colorado AI Act deployer obligations, and BIPA implications of any biometric component. Re-derive.\n\nA second failure mode: the footprint was set before a regime that now matters existed (or took effect). If re-derivation surfaces a regime not in the footprint, flag it in the output's recommendation section, cite the authority, and recommend updating the footprint.\n\nFor each regime in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` → `## Regulatory footprint` that applies to this system — **plus any regime surfaced by the re-derivation above** — research the currently operative risk classification framework and determine where the system lands.\n\nResearch tasks:\n- What is the regime's own tier taxonomy (e.g., prohibited \u002F high-risk \u002F limited \u002F minimal, or the regime's equivalent)?\n- What are the criteria for each tier? Cite primary sources with pinpoint references.\n- Which tier does this system fall into given its function, affected parties, and decision consequentiality?\n- Are there prohibited practices the system might touch? Treat any possible match as critical — flag immediately.\n- Are there transparency obligations that apply regardless of tier (disclosure that a user is interacting with AI, labeling of AI-generated content, notice to people subject to automated decisions)?\n- If the company is a builder providing a general-purpose or foundation model, what provider-level obligations apply (technical documentation, training data transparency, copyright compliance, systemic-risk testing)?\n- **Does any regime in the footprint require a separate fundamental-rights impact assessment (FRIA)?** EU AI Act Art. 27 requires a FRIA for certain deployers of high-risk AI systems (public bodies and private entities providing public services, plus certain creditworthiness and insurance-risk-assessment use cases). Check each regime for an equivalent fundamental-rights or human-rights impact assessment that is a distinct deliverable from this AIA. If a FRIA (or regime equivalent) is required, flag it as a separate deliverable in the recommendation and conditions — do not treat this AIA as a substitute.\n\nDon't assume internal-only systems are out of scope — most regimes treat employee data as personal data and employee monitoring as consequential. Verify the specific rule.\n\n**Provider-vs-deployer split (when `AI role: Both`).** If `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` → `## Company profile` → `AI role` is `Both` (the company is both a provider\u002Fbuilder and a deployer), Section 6 MUST include a provider-vs-deployer mapping table per regime. Most regimes impose materially different obligations on providers (or builders) versus deployers (or users) — collapsing them into one undifferentiated list misses obligations and conflates risks. Do not combine provider and deployer obligations into a single section. Produce, per regime:\n\n| Obligation | As provider | As deployer |\n|---|---|---|\n| [specific obligation, pinpoint cite] | [what applies \u002F does not apply \u002F with what carve-outs] | [what applies \u002F does not apply \u002F with what carve-outs] |\n\n**If a high-risk or equivalent classification applies:**\nFlag in the assessment, citing the specific provision and regime. Note that this AIA documents the internal review but does not substitute for any formal conformity assessment the regime requires. Recommend external legal review before deployment in the affected jurisdiction.\n\nCapture the classification and the cited authority in the assessment output.\n\n---\n\n## Step 4: Write the assessment\n\n**Use the seed structure from `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`.** If none was captured, use this default:\n\n```markdown\n[WORK-PRODUCT HEADER — per plugin config ## Outputs — differs by role; see `## Who's using this`]\n\n# AI Impact Assessment: [System\u002FFeature Name]\n\n**Prepared by:** [name] | **Date:** [date] | **Status:** DRAFT \u002F APPROVED\n**System owner:** [name] | **AI governance reviewer:** [name]\n**Governance tier:** [Standard \u002F Elevated \u002F High]\n**Track:** [Fast track \u002F Full assessment]\n\n---\n\n## Executive summary\n\n[Two sentences: what this AI does and whether it's okay to deploy. E.g., \"This\nsystem uses a third-party LLM to draft initial responses to customer support tickets\nbefore human agent review. Processing is consistent with the company's AI policy;\nthree conditions required before production deployment.\"]\n\n**Overall risk:** 🟢 Low \u002F 🟡 Medium \u002F 🟠 High \u002F 🔴 Very high\n\n---\n\n## 1. System description\n\n**What it does:** [plain English — not marketing]\n**Model \u002F vendor:** [who's providing the AI]\n**Deployment mode:** [Assistive \u002F Augmentative \u002F Automated]\n**Output type:** [text \u002F score \u002F classification \u002F recommendation \u002F action]\n**Status:** [Not started \u002F Pilot \u002F Production]\n\n---\n\n## 2. Affected parties\n\n**Who it acts on:** [employees \u002F customers \u002F third parties]\n**Scale:** [how many people, how often]\n**Harm if wrong:** [most realistic worst case — specific, not generic]\n**Vulnerable groups in scope:** [yes — [who] \u002F no]\n\n---\n\n## 3. Data inputs\n\n**Data categories used:** [specific fields, not \"user data\"]\n**Personal data:** [yes — [whose] \u002F no]\n**Data leaves perimeter?** [yes — to [vendor] \u002F no]\n**Model training:** [company data used \u002F foundation model \u002F fine-tuned on [dataset]]\n\n---\n\n## 4. Decision-making and oversight\n\n**Human in the loop:** [Always \u002F Nominally (rubber-stamp risk) \u002F No]\n**Override mechanism:** [how a human can intervene or correct]\n**Appeals \u002F correction for affected parties:** [yes — [how] \u002F no]\n**Named owner:** [name or role]\n\n---\n\n## 5. Accuracy and bias\n\n**Error rate:** [known \u002F estimated \u002F untested]\n**Failure mode:** [what happens when it's wrong — surfaced? logged? corrected?]\n**Bias testing:** [done — [results] \u002F not done \u002F not applicable]\n\n---\n\n## 6. Regulatory classification\n\n*[One subsection per regime in the regulatory footprint that applies to this system.]*\n\n**Regime:** [name]\n**Classification under this regime:** [tier, with pinpoint citation to the controlling provision]\n**Prohibited practices triggered:** [none identified \u002F [specific provision and why]]\n**Applicable obligations:** [researched list with citations — transparency, documentation, human oversight, testing, registration, etc.]\n**Fundamental-rights impact assessment required?** [Yes — e.g., EU AI Act Art. 27 FRIA applies \u002F regime equivalent \u002F No \u002F Not applicable. If yes, this is a separate deliverable, not subsumed by this AIA.]\n**Effective \u002F enforcement date:** [date(s)]\n**Ambiguity or open interpretation:** [flag anything not yet settled]\n\n**Provider-vs-deployer obligation split (required if `AI role: Both`):**\n\n| Obligation | As provider | As deployer |\n|---|---|---|\n| [specific obligation + pinpoint cite] | [what applies \u002F does not apply] | [what applies \u002F does not apply] |\n\n---\n\n## 7. AI policy consistency\n\n| Policy commitment | Consistent? | Notes |\n|---|---|---|\n| [commitment from `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` AI policy section] | 🟢 \u002F 🟡 \u002F 🟠 \u002F 🔴 | |\n\n[If any item is 🟡 or worse: policy update needed before deployment, or design needs to change.\nOne of them has to change — not both flagged and left open.]\n\n---\n\n## 8. Risks and mitigations\n\n| # | Risk | Likelihood | Impact | Mitigation | Status | Owner |\n|---|---|---|---|---|---|---|\n| 1 | [specific risk tied to this design — not \"AI hallucination\" generically] | L\u002FM\u002FH | L\u002FM\u002FH | [specific control] | Done \u002F Planned \u002F Gap | [name] |\n\n**Residual risk after mitigations:** [assessment]\n\n---\n\n## 9. Recommendation\n\n**[APPROVED \u002F APPROVED WITH CONDITIONS \u002F CHANGES REQUIRED \u002F NOT APPROVED]**\n\n**Conditions (if any):**\n- [ ] [specific action before deployment — owner, deadline]\n\n**Privacy review required?** [Yes — run `\u002Fprivacy-legal:pia-generation`, if the plugin is installed \u002F\nNo]\n\n**Sign-off:** [name, date]\n\n---\n\n## Cite check\n\nRegulatory citations in Section 6 (and anywhere else) were generated by an AI model and have not been verified against primary sources. Before the assessment is certified or relied on, run a verification pass against a legal research tool (Westlaw, EUR-Lex, or your firm's platform) for each cited provision — confirm the pinpoint, currency, and any delegated or implementing acts. The AI regulatory landscape shifts quickly; verify before advising. Source tags on each citation (e.g., `[EUR-Lex]`, `[web search — verify]`) show where it came from; `verify` tags carry higher fabrication risk and should be checked first.\n```\n\n**Before certifying the AIA (the Sign-off step, marking Status: APPROVED):** Read `## Who's using this` in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`. If the Role is Non-lawyer:\n\n> Certifying this AIA has legal consequences — it becomes the record the company relies on if a regulator or affected party asks how this use case was assessed. Have you reviewed this with an attorney? If yes, proceed. If no, here's a brief to bring to them:\n>\n> [Generate a 1-page summary: the system, the regulatory classification, the risks identified, the mitigations in place, residual risk, open questions, what to ask the attorney before certifying.]\n>\n> If you need to find an attorney, solicitor, barrister, or other authorised legal professional: your professional regulator's referral service is the fastest starting point (state bar in the US, SRA\u002FBar Standards Board in England & Wales, Law Society in Scotland\u002FNI\u002FIreland\u002FCanada\u002FAustralia, or your jurisdiction's equivalent).\n\nDo not proceed past this gate without an explicit yes. DRAFT assessments for attorney review do not require the gate — certification does.\n\n---\n\n## Risk quality standards\n\nSame standard as the PIA skill — risks must be **specific and tied to the design**.\n\n| Bad risk | Why bad | Better |\n|---|---|---|\n| \"AI hallucination\" | Applies to every LLM; says nothing | \"Model may generate plausible but incorrect legal citations — support agents have no current verification step before sending to customers\" |\n| \"Bias\" | Too vague | \"Résumé scoring model trained on historical hires; if historical cohort was demographically homogeneous, underrepresented candidates may be systematically scored lower\" |\n| \"Vendor risk\" | Circular | \"OpenAI's terms permit training on API inputs by default; unless the opt-out is confirmed in the agreement, customer support messages may be used to train the model\" |\n\nAim for 2-5 real risks, not 12 padded ones.\n\n---\n\n## AI policy diff\n\nEvery assessment should cross-check against the AI policy commitments in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md`.\nCommon drift:\n\n- Policy prohibits AI use in [category] — this use case is that category. Stop.\n- Policy requires human review — this deployment has no human step. Design needs to change.\n- Policy requires disclosure to affected parties — disclosure mechanism hasn't been built.\n- Approved vendor list exists — this vendor isn't on it. Procurement step required.\n\nFlag every mismatch. One of them has to change before deployment.\n\n---\n\n## Handoffs\n\n- **To product \u002F engineering:** Conditions list with owners and deadlines. Not\n  \"add oversight\" — \"add a human review step before any automated email is sent,\n  owner: [product lead], before launch.\"\n- **To privacy:** If personal data is involved, flag: \"Run `\u002Fprivacy-legal:pia-generation [system name]` in parallel, if the plugin is installed — the AIA doesn't substitute for a PIA.\"\n- **To vendor-ai-review:** If a new vendor is involved, flag: \"If there's no AI addendum reviewed for [vendor], run `\u002Fai-governance-legal:vendor-ai-review` before production.\"\n- **To reg-gap-analysis:** If new regulatory obligations emerged (EU AI Act high-risk, new sector rule), that skill tracks the gap.\n\n---\n\n## Close with the next-steps decision tree\n\nEnd with the next-steps decision tree per CLAUDE.md `## Outputs`. Customize the options to what this skill just produced — the five default branches (draft the X, escalate, get more facts, watch and wait, something else) are a starting point, not a lock-in. The tree is the output; the lawyer picks.\n\n## What this skill does not do\n\n- It doesn't approve the deployment. A human signs the assessment.\n- It doesn't constitute any regulatory conformity assessment — where a regime (e.g., EU AI Act) requires a formal conformity assessment, that is a separate exercise requiring external legal review and technical documentation beyond what's here.\n- It doesn't design the mitigations. It describes what needs mitigating; engineering\n  designs the fix.\n- It doesn't substitute for a PIA when personal data is involved. Run both.\n",{"data":38,"body":40},{"name":4,"description":6,"argument-hint":39},"[describe the use case or system, or pass a triage result]",{"type":41,"children":42},"root",[43,51,107,119,123,130,213,216,222,227,232,238,257,281,300,325,328,334,345,353,388,393,402,405,411,439,444,657,681,686,689,695,700,707,730,736,754,760,783,789,812,818,836,842,847,887,892,895,901,923,928,980,985,990,1015,1020,1063,1068,1113,1172,1182,1187,1190,1196,1212,3424,3447,3468,3473,3476,3482,3493,3574,3579,3582,3588,3600,3630,3635,3638,3644,3716,3719,3725,3738,3744,3767],{"type":44,"tag":45,"props":46,"children":47},"element","h1",{"id":4},[48],{"type":49,"value":50},"text","\u002Faia-generation",{"type":44,"tag":52,"props":53,"children":54},"ol",{},[55,70,75,80,85,90,102],{"type":44,"tag":56,"props":57,"children":58},"li",{},[59,61,68],{"type":49,"value":60},"Read ",{"type":44,"tag":62,"props":63,"children":65},"code",{"className":64},[],[66],{"type":49,"value":67},"~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md",{"type":49,"value":69},". Confirm impact assessment house style is populated.",{"type":44,"tag":56,"props":71,"children":72},{},[73],{"type":49,"value":74},"Determine risk track (fast or full) from governance tier and use case characteristics, using the framework below.",{"type":44,"tag":56,"props":76,"children":77},{},[78],{"type":49,"value":79},"Run intake — conversational, not a form.",{"type":44,"tag":56,"props":81,"children":82},{},[83],{"type":49,"value":84},"Regulatory classification for each regime in the footprint — research tier, prohibited-practice exposure, and applicable obligations; cite primary sources.",{"type":44,"tag":56,"props":86,"children":87},{},[88],{"type":49,"value":89},"Write assessment in house style (from seed doc, or default if none captured).",{"type":44,"tag":56,"props":91,"children":92},{},[93,95,100],{"type":49,"value":94},"Policy diff against ",{"type":44,"tag":62,"props":96,"children":98},{"className":97},[],[99],{"type":49,"value":67},{"type":49,"value":101}," AI policy commitments.",{"type":44,"tag":56,"props":103,"children":104},{},[105],{"type":49,"value":106},"Output: assessment doc + conditions list + handoff flags (privacy PIA, vendor review if needed).",{"type":44,"tag":108,"props":109,"children":113},"pre",{"className":110,"code":112,"language":49},[111],"language-text","\u002Fai-governance-legal:aia-generation \"AI résumé screening for HR\"\n",[114],{"type":44,"tag":62,"props":115,"children":117},{"__ignoreMap":116},"",[118],{"type":49,"value":112},{"type":44,"tag":120,"props":121,"children":122},"hr",{},[],{"type":44,"tag":124,"props":125,"children":127},"h2",{"id":126},"matter-context",[128],{"type":49,"value":129},"Matter context",{"type":44,"tag":131,"props":132,"children":133},"p",{},[134,140,142,148,150,156,158,164,166,172,174,180,182,188,190,196,198,204,205,211],{"type":44,"tag":135,"props":136,"children":137},"strong",{},[138],{"type":49,"value":139},"Matter context.",{"type":49,"value":141}," Check ",{"type":44,"tag":62,"props":143,"children":145},{"className":144},[],[146],{"type":49,"value":147},"## Matter workspaces",{"type":49,"value":149}," in the practice-level CLAUDE.md. If ",{"type":44,"tag":62,"props":151,"children":153},{"className":152},[],[154],{"type":49,"value":155},"Enabled",{"type":49,"value":157}," is ",{"type":44,"tag":62,"props":159,"children":161},{"className":160},[],[162],{"type":49,"value":163},"✗",{"type":49,"value":165}," (the default for in-house users), skip the rest of this paragraph — skills use practice-level context and the matter machinery is invisible. If enabled and there is no active matter, ask: \"Which matter is this for? Run ",{"type":44,"tag":62,"props":167,"children":169},{"className":168},[],[170],{"type":49,"value":171},"\u002Fai-governance-legal:matter-workspace switch \u003Cslug>",{"type":49,"value":173}," or say ",{"type":44,"tag":62,"props":175,"children":177},{"className":176},[],[178],{"type":49,"value":179},"practice-level",{"type":49,"value":181},".\" Load the active matter's ",{"type":44,"tag":62,"props":183,"children":185},{"className":184},[],[186],{"type":49,"value":187},"matter.md",{"type":49,"value":189}," for matter-specific context and overrides. Write outputs to the matter folder at ",{"type":44,"tag":62,"props":191,"children":193},{"className":192},[],[194],{"type":49,"value":195},"~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002Fmatters\u002F\u003Cmatter-slug>\u002F",{"type":49,"value":197},". Never read another matter's files unless ",{"type":44,"tag":62,"props":199,"children":201},{"className":200},[],[202],{"type":49,"value":203},"Cross-matter context",{"type":49,"value":157},{"type":44,"tag":62,"props":206,"children":208},{"className":207},[],[209],{"type":49,"value":210},"on",{"type":49,"value":212},".",{"type":44,"tag":120,"props":214,"children":215},{},[],{"type":44,"tag":124,"props":217,"children":219},{"id":218},"purpose",[220],{"type":49,"value":221},"Purpose",{"type":44,"tag":131,"props":223,"children":224},{},[225],{"type":49,"value":226},"An AI impact assessment is a documented decision, not a form. It answers: what\ndoes this AI system do, how does it reach its outputs, who's affected if it's\nwrong, what's the oversight, and is it okay to deploy. This skill structures that\nconversation and writes the output in this team's format — the one learned from the\nseed impact assessment during cold-start.",{"type":44,"tag":131,"props":228,"children":229},{},[230],{"type":49,"value":231},"An AI impact assessment is not the same as a PIA. A PIA asks whether personal data\nis handled lawfully. An AIA asks whether the AI system is designed and deployed\nresponsibly. They often need to happen in parallel; they're not substitutes.",{"type":44,"tag":124,"props":233,"children":235},{"id":234},"load-house-style",[236],{"type":49,"value":237},"Load house style",{"type":44,"tag":131,"props":239,"children":240},{},[241,242,247,249,255],{"type":49,"value":60},{"type":44,"tag":62,"props":243,"children":245},{"className":244},[],[246],{"type":49,"value":67},{"type":49,"value":248}," → ",{"type":44,"tag":62,"props":250,"children":252},{"className":251},[],[253],{"type":49,"value":254},"## Impact assessment house style",{"type":49,"value":256},". That has:",{"type":44,"tag":258,"props":259,"children":260},"ul",{},[261,266,271,276],{"type":44,"tag":56,"props":262,"children":263},{},[264],{"type":49,"value":265},"What triggers an impact assessment at this company",{"type":44,"tag":56,"props":267,"children":268},{},[269],{"type":49,"value":270},"The structure template extracted from the seed assessment",{"type":44,"tag":56,"props":272,"children":273},{},[274],{"type":49,"value":275},"Typical depth",{"type":44,"tag":56,"props":277,"children":278},{},[279],{"type":49,"value":280},"Who signs off",{"type":44,"tag":131,"props":282,"children":283},{},[284,286,291,293,298],{"type":49,"value":285},"If the seed structure is in ",{"type":44,"tag":62,"props":287,"children":289},{"className":288},[],[290],{"type":49,"value":67},{"type":49,"value":292},", ",{"type":44,"tag":135,"props":294,"children":295},{},[296],{"type":49,"value":297},"use it",{"type":49,"value":299},". The point is that this assessment\nlooks like the other assessments this team produces.",{"type":44,"tag":131,"props":301,"children":302},{},[303,308,310,316,318,323],{"type":44,"tag":135,"props":304,"children":305},{},[306],{"type":49,"value":307},"Jurisdictional scope.",{"type":49,"value":309}," This assessment applies the regulatory regimes listed in ",{"type":44,"tag":62,"props":311,"children":313},{"className":312},[],[314],{"type":49,"value":315},"## Regulatory footprint",{"type":49,"value":317}," in ",{"type":44,"tag":62,"props":319,"children":321},{"className":320},[],[322],{"type":49,"value":67},{"type":49,"value":324},". AI legal rules, risk classifications, and deployment obligations vary materially by jurisdiction and are moving fast. If this system is (or will be) deployed outside that footprint, or if a choice-of-law question is in play, this analysis may not apply as written — re-run or expand the footprint.",{"type":44,"tag":120,"props":326,"children":327},{},[],{"type":44,"tag":124,"props":329,"children":331},{"id":330},"step-0-is-an-impact-assessment-needed",[332],{"type":49,"value":333},"Step 0: Is an impact assessment needed?",{"type":44,"tag":131,"props":335,"children":336},{},[337,339,344],{"type":49,"value":338},"Check the trigger criteria in ",{"type":44,"tag":62,"props":340,"children":342},{"className":341},[],[343],{"type":49,"value":67},{"type":49,"value":212},{"type":44,"tag":131,"props":346,"children":347},{},[348],{"type":44,"tag":135,"props":349,"children":350},{},[351],{"type":49,"value":352},"Also check these regardless:",{"type":44,"tag":258,"props":354,"children":355},{},[356,361,366,371,376],{"type":44,"tag":56,"props":357,"children":358},{},[359],{"type":49,"value":360},"Does this AI make or materially influence a decision affecting a person (employment,\ncredit, access, pricing, content moderation)?",{"type":44,"tag":56,"props":362,"children":363},{},[364],{"type":49,"value":365},"Does this AI process personal data about individuals?",{"type":44,"tag":56,"props":367,"children":368},{},[369],{"type":49,"value":370},"Is this a customer-facing AI system rather than purely internal?",{"type":44,"tag":56,"props":372,"children":373},{},[374],{"type":49,"value":375},"Does this AI use a third-party model where the company is the deployer?",{"type":44,"tag":56,"props":377,"children":378},{},[379,381,386],{"type":49,"value":380},"Is the use case in the elevated or high governance tier per ",{"type":44,"tag":62,"props":382,"children":384},{"className":383},[],[385],{"type":49,"value":67},{"type":49,"value":387},"?",{"type":44,"tag":131,"props":389,"children":390},{},[391],{"type":49,"value":392},"If none of the above and the house trigger isn't met:",{"type":44,"tag":394,"props":395,"children":396},"blockquote",{},[397],{"type":44,"tag":131,"props":398,"children":399},{},[400],{"type":49,"value":401},"\"Doesn't look like this needs a full impact assessment. Here's a one-paragraph\nrecord for the file explaining why — in case anyone asks later.\"",{"type":44,"tag":120,"props":403,"children":404},{},[],{"type":44,"tag":124,"props":406,"children":408},{"id":407},"step-1-risk-track",[409],{"type":49,"value":410},"Step 1: Risk track",{"type":44,"tag":131,"props":412,"children":413},{},[414,416,421,423,429,431,437],{"type":49,"value":415},"Before intake, determine which track to run. The tier definitions and the fast-track criteria come from ",{"type":44,"tag":62,"props":417,"children":419},{"className":418},[],[420],{"type":49,"value":67},{"type":49,"value":422}," (",{"type":44,"tag":62,"props":424,"children":426},{"className":425},[],[427],{"type":49,"value":428},"## Use case registry",{"type":49,"value":430}," and ",{"type":44,"tag":62,"props":432,"children":434},{"className":433},[],[435],{"type":49,"value":436},"## Governance tiers",{"type":49,"value":438},"), not from any hardcoded regime-specific framework.",{"type":44,"tag":131,"props":440,"children":441},{},[442],{"type":49,"value":443},"Research the applicable risk classification framework for each regime in the user's regulatory footprint. Many regimes distinguish by risk tier, affected population, and decision consequentiality — research the specific criteria. Note that most regimes treat employee data as personal data and employee monitoring as consequential; don't assume internal-only systems are out of scope.",{"type":44,"tag":394,"props":445,"children":446},{},[447,487,497,533,575,634,645],{"type":44,"tag":131,"props":448,"children":449},{},[450,455,457,463,465,470,472,477,479,485],{"type":44,"tag":135,"props":451,"children":452},{},[453],{"type":49,"value":454},"No silent supplement.",{"type":49,"value":456}," If a research query to the configured legal research tool (Westlaw, EUR-Lex, regulator sites, or firm platform) returns few or no results for a regime's risk tiers or triggers, report what was found and stop. Do NOT fill the gap from web search or model knowledge without asking. Say: \"The search returned ",{"type":44,"tag":458,"props":459,"children":460},"span",{},[461],{"type":49,"value":462},"N",{"type":49,"value":464}," results from ",{"type":44,"tag":458,"props":466,"children":467},{},[468],{"type":49,"value":469},"tool",{"type":49,"value":471},". Coverage appears thin for ",{"type":44,"tag":458,"props":473,"children":474},{},[475],{"type":49,"value":476},"regime \u002F topic",{"type":49,"value":478},". Options: (1) broaden the search query, (2) try a different research tool, (3) search the web — results will be tagged ",{"type":44,"tag":62,"props":480,"children":482},{"className":481},[],[483],{"type":49,"value":484},"[web search — verify]",{"type":49,"value":486}," and should be checked against the issuing authority before relying, or (4) flag as unverified and stop. Which would you like?\" A lawyer decides whether to accept lower-confidence sources.",{"type":44,"tag":131,"props":488,"children":489},{},[490,495],{"type":44,"tag":135,"props":491,"children":492},{},[493],{"type":49,"value":494},"Source attribution tiering.",{"type":49,"value":496}," Tag every citation in the AIA — regulatory text, delegated acts, guidance, standards — with its source. For model-knowledge citations, use one of three tiers rather than a single blanket \"verify\" tag:",{"type":44,"tag":258,"props":498,"children":499},{},[500,511,522],{"type":44,"tag":56,"props":501,"children":502},{},[503,509],{"type":44,"tag":62,"props":504,"children":506},{"className":505},[],[507],{"type":49,"value":508},"[settled]",{"type":49,"value":510}," — stable, well-known statutory and regulatory references unlikely to have changed (e.g., GDPR Art. 22 as a concept, the existence of Regulation (EU) 2024\u002F1689 as the EU AI Act). Still verify before certifying, but lower priority.",{"type":44,"tag":56,"props":512,"children":513},{},[514,520],{"type":44,"tag":62,"props":515,"children":517},{"className":516},[],[518],{"type":49,"value":519},"[verify]",{"type":49,"value":521}," — model-knowledge citations that are real but should be verified: specific delegated \u002F implementing acts, regulator guidance, NYC DCWP rules, Colorado AI Act provisions, harmonized standards, effective dates, EEOC guidance, and anything post-2023.",{"type":44,"tag":56,"props":523,"children":524},{},[525,531],{"type":44,"tag":62,"props":526,"children":528},{"className":527},[],[529],{"type":49,"value":530},"[verify-pinpoint]",{"type":49,"value":532}," — pinpoint citations (specific EU AI Act article numbers, annex references, Colorado AI Act subsections, NYC LL 144 rule sections, sub-paragraph letters) carry the highest fabrication risk and should ALWAYS be verified against a primary source. EU AI Act article numbers in particular shifted during consolidation; every pinpoint cite to the Act should be verified against the Official Journal text.",{"type":44,"tag":131,"props":534,"children":535},{},[536,538,544,545,551,552,558,560,565,567,573],{"type":49,"value":537},"Tool-retrieved citations keep their source tag (",{"type":44,"tag":62,"props":539,"children":541},{"className":540},[],[542],{"type":49,"value":543},"[Westlaw]",{"type":49,"value":292},{"type":44,"tag":62,"props":546,"children":548},{"className":547},[],[549],{"type":49,"value":550},"[EUR-Lex]",{"type":49,"value":292},{"type":44,"tag":62,"props":553,"children":555},{"className":554},[],[556],{"type":49,"value":557},"[regulator site]",{"type":49,"value":559},", or the MCP tool name); web-search citations remain ",{"type":44,"tag":62,"props":561,"children":563},{"className":562},[],[564],{"type":49,"value":484},{"type":49,"value":566},"; user-supplied citations remain ",{"type":44,"tag":62,"props":568,"children":570},{"className":569},[],[571],{"type":49,"value":572},"[user provided]",{"type":49,"value":574},". The tiering surfaces the real verification work — a reader who verifies everything verifies nothing. Never strip or collapse the tags.",{"type":44,"tag":131,"props":576,"children":577},{},[578,583,585,590,592,597,599,605,606,611,613,618,620,625,627,632],{"type":44,"tag":135,"props":579,"children":580},{},[581],{"type":49,"value":582},"For non-lawyer users, uncertain dates go in a confirm-list, not inline.",{"type":49,"value":584}," A ",{"type":44,"tag":62,"props":586,"children":588},{"className":587},[],[589],{"type":49,"value":519},{"type":49,"value":591}," tag on \"effective February 1, 2026\" reads as \"effective February 1, 2026\" to a CISO who doesn't know what ",{"type":44,"tag":62,"props":593,"children":595},{"className":594},[],[596],{"type":49,"value":519},{"type":49,"value":598}," means. Read ",{"type":44,"tag":62,"props":600,"children":602},{"className":601},[],[603],{"type":49,"value":604},"## Who's using this",{"type":49,"value":317},{"type":44,"tag":62,"props":607,"children":609},{"className":608},[],[610],{"type":49,"value":67},{"type":49,"value":612},". If Role is ",{"type":44,"tag":135,"props":614,"children":615},{},[616],{"type":49,"value":617},"Non-lawyer",{"type":49,"value":619}," and a date, deadline, phase-in, threshold, or effective-date assertion is uncertain (would carry ",{"type":44,"tag":62,"props":621,"children":623},{"className":622},[],[624],{"type":49,"value":519},{"type":49,"value":626}," or ",{"type":44,"tag":62,"props":628,"children":630},{"className":629},[],[631],{"type":49,"value":530},{"type":49,"value":633}," if inline), replace the inline assertion with \"effective date: confirm with counsel\" (or \"threshold: confirm with counsel\", etc.) and collect all uncertain assertions in a final AIA section titled:",{"type":44,"tag":394,"props":635,"children":636},{},[637],{"type":44,"tag":131,"props":638,"children":639},{},[640],{"type":44,"tag":135,"props":641,"children":642},{},[643],{"type":49,"value":644},"Things I'm not certain about — ask your attorney to confirm before relying on this:",{"type":44,"tag":131,"props":646,"children":647},{},[648,650,655],{"type":49,"value":649},"List each uncertain item there with (1) what I said, (2) what I'm uncertain about, (3) why it matters to the assessment. This prevents a non-lawyer reader from mistaking a flagged best-guess for a checked fact. Lawyer-role users get the inline ",{"type":44,"tag":62,"props":651,"children":653},{"className":652},[],[654],{"type":49,"value":519},{"type":49,"value":656}," treatment — they know what the tag means.",{"type":44,"tag":131,"props":658,"children":659},{},[660,665,667,672,674,679],{"type":44,"tag":135,"props":661,"children":662},{},[663],{"type":49,"value":664},"Fast track vs. full assessment:",{"type":49,"value":666}," ",{"type":44,"tag":62,"props":668,"children":670},{"className":669},[],[671],{"type":49,"value":67},{"type":49,"value":673}," defines what qualifies for abbreviated treatment. If ",{"type":44,"tag":62,"props":675,"children":677},{"className":676},[],[678],{"type":49,"value":67},{"type":49,"value":680}," doesn't define fast-track criteria, default to full assessment and ask the user what criteria they want captured for next time.",{"type":44,"tag":131,"props":682,"children":683},{},[684],{"type":49,"value":685},"If in doubt, run the full assessment. A fast track that turns out to be wrong\nis worse than a thorough assessment on something low-risk.",{"type":44,"tag":120,"props":687,"children":688},{},[],{"type":44,"tag":124,"props":690,"children":692},{"id":691},"step-2-intake",[693],{"type":49,"value":694},"Step 2: Intake",{"type":44,"tag":131,"props":696,"children":697},{},[698],{"type":49,"value":699},"Before writing anything, get answers to these. Conversational is fine — this\nis not a form to send them.",{"type":44,"tag":701,"props":702,"children":704},"h3",{"id":703},"the-system",[705],{"type":49,"value":706},"The system",{"type":44,"tag":258,"props":708,"children":709},{},[710,715,720,725],{"type":44,"tag":56,"props":711,"children":712},{},[713],{"type":49,"value":714},"What does the AI do? Describe it in plain language, not marketing copy.",{"type":44,"tag":56,"props":716,"children":717},{},[718],{"type":49,"value":719},"Which model or vendor is powering it? Fine-tuned or off-the-shelf?",{"type":44,"tag":56,"props":721,"children":722},{},[723],{"type":49,"value":724},"Where does it sit in the workflow — is it assistive (human reviews output),\naugmentative (human can override but usually doesn't), or automated (no human\nin the loop)?",{"type":44,"tag":56,"props":726,"children":727},{},[728],{"type":49,"value":729},"What's the output — generated text, a score, a classification, a recommendation,\nan action?",{"type":44,"tag":701,"props":731,"children":733},{"id":732},"whos-affected",[734],{"type":49,"value":735},"Who's affected",{"type":44,"tag":258,"props":737,"children":738},{},[739,744,749],{"type":44,"tag":56,"props":740,"children":741},{},[742],{"type":49,"value":743},"Who does the AI's output act on — employees, customers, third parties?",{"type":44,"tag":56,"props":745,"children":746},{},[747],{"type":49,"value":748},"If the AI produces an error (false positive, false negative, hallucination), who\nbears the harm and what's the worst realistic case?",{"type":44,"tag":56,"props":750,"children":751},{},[752],{"type":49,"value":753},"Are any vulnerable groups disproportionately in scope — minors, job applicants,\npeople in financial distress, patients?",{"type":44,"tag":701,"props":755,"children":757},{"id":756},"inputs-and-data",[758],{"type":49,"value":759},"Inputs and data",{"type":44,"tag":258,"props":761,"children":762},{},[763,768,773,778],{"type":44,"tag":56,"props":764,"children":765},{},[766],{"type":49,"value":767},"What data does the AI take in?",{"type":44,"tag":56,"props":769,"children":770},{},[771],{"type":49,"value":772},"Does it take in personal data? Whose?",{"type":44,"tag":56,"props":774,"children":775},{},[776],{"type":49,"value":777},"Was the model trained on data from this company, or is it a foundation model\nwith no company-specific training?",{"type":44,"tag":56,"props":779,"children":780},{},[781],{"type":49,"value":782},"Where does input data go — does it leave the perimeter to a third-party model\nAPI?",{"type":44,"tag":701,"props":784,"children":786},{"id":785},"decisions-and-oversight",[787],{"type":49,"value":788},"Decisions and oversight",{"type":44,"tag":258,"props":790,"children":791},{},[792,797,802,807],{"type":44,"tag":56,"props":793,"children":794},{},[795],{"type":49,"value":796},"Does the AI output trigger an action automatically, or does a human decide what\nto do with the output?",{"type":44,"tag":56,"props":798,"children":799},{},[800],{"type":49,"value":801},"If there's human review: how often does the human actually change the AI's output?\n(If the answer is \"rarely\" — the human isn't really reviewing; they're rubber-stamping.)",{"type":44,"tag":56,"props":803,"children":804},{},[805],{"type":49,"value":806},"Is there an appeals or correction process for people affected by the AI's outputs?",{"type":44,"tag":56,"props":808,"children":809},{},[810],{"type":49,"value":811},"Who is accountable for the AI system's outputs — is there a named owner?",{"type":44,"tag":701,"props":813,"children":815},{"id":814},"accuracy-and-failure",[816],{"type":49,"value":817},"Accuracy and failure",{"type":44,"tag":258,"props":819,"children":820},{},[821,826,831],{"type":44,"tag":56,"props":822,"children":823},{},[824],{"type":49,"value":825},"What's the known or estimated error rate? What testing has been done?",{"type":44,"tag":56,"props":827,"children":828},{},[829],{"type":49,"value":830},"What happens when the AI is wrong — is the error surfaced, logged, corrected?",{"type":44,"tag":56,"props":832,"children":833},{},[834],{"type":49,"value":835},"Has bias testing been done? Against what demographic groups?",{"type":44,"tag":701,"props":837,"children":839},{"id":838},"deployment-stage-and-scale",[840],{"type":49,"value":841},"Deployment stage and scale",{"type":44,"tag":131,"props":843,"children":844},{},[845],{"type":49,"value":846},"Ask:",{"type":44,"tag":258,"props":848,"children":849},{},[850,860,877],{"type":44,"tag":56,"props":851,"children":852},{},[853,858],{"type":44,"tag":135,"props":854,"children":855},{},[856],{"type":49,"value":857},"Stage:",{"type":49,"value":859}," \"Is this system (a) proposed and not yet built, (b) in pilot, (c) live in production, or (d) live and scaled?\"",{"type":44,"tag":56,"props":861,"children":862},{},[863,868,870,875],{"type":44,"tag":135,"props":864,"children":865},{},[866],{"type":49,"value":867},"Scale:",{"type":49,"value":869}," \"Roughly how many individuals are affected per ",{"type":44,"tag":458,"props":871,"children":872},{},[873],{"type":49,"value":874},"month\u002Fyear",{"type":49,"value":876},"? How long has it been running?\"",{"type":44,"tag":56,"props":878,"children":879},{},[880,885],{"type":44,"tag":135,"props":881,"children":882},{},[883],{"type":49,"value":884},"History:",{"type":49,"value":886}," \"Has it been assessed before? Has it produced decisions that were challenged, appealed, or reversed?\"",{"type":44,"tag":131,"props":888,"children":889},{},[890],{"type":49,"value":891},"Stage changes the assessment: a proposed system gets a design review (can we build it safely?). A pilot gets a design review plus a \"before you scale\" gate. A live system gets a retrospective impact check (has it caused harm?) AND a go-forward review. A live-and-scaled system gets all of the above plus a remediation plan if issues are found, because you can't just turn it off.",{"type":44,"tag":120,"props":893,"children":894},{},[],{"type":44,"tag":124,"props":896,"children":898},{"id":897},"step-3-regulatory-classification",[899],{"type":49,"value":900},"Step 3: Regulatory classification",{"type":44,"tag":131,"props":902,"children":903},{},[904,909,911,916,918],{"type":44,"tag":135,"props":905,"children":906},{},[907],{"type":49,"value":908},"Step 3 pre-check — footprint freshness.",{"type":49,"value":910}," Before iterating over the captured ",{"type":44,"tag":62,"props":912,"children":914},{"className":913},[],[915],{"type":49,"value":315},{"type":49,"value":917},", compare the use case's affected population and decision type (from Step 2) against the footprint as written. The footprint was set at cold-start, based on the company's operating posture at that moment. If the use case introduces an affected population (e.g., children, employees in a new state, EU data subjects) or a decision type (e.g., hiring, creditworthiness, health diagnosis, law enforcement, critical infrastructure) that the footprint does not contemplate, ",{"type":44,"tag":135,"props":919,"children":920},{},[921],{"type":49,"value":922},"re-derive the applicable regimes rather than iterating over the stale list.",{"type":44,"tag":131,"props":924,"children":925},{},[926],{"type":49,"value":927},"Say to the user:",{"type":44,"tag":394,"props":929,"children":930},{},[931],{"type":44,"tag":131,"props":932,"children":933},{},[934,936,941,943,951,953,964,966,971,973,978],{"type":49,"value":935},"\"The practice profile's regulatory footprint was set for ",{"type":44,"tag":458,"props":937,"children":938},{},[939],{"type":49,"value":940},"affected populations \u002F decision types captured at cold-start",{"type":49,"value":942},". This use case affects ",{"type":44,"tag":135,"props":944,"children":945},{},[946],{"type":44,"tag":458,"props":947,"children":948},{},[949],{"type":49,"value":950},"new population or decision type — e.g., employees in Colorado, minors under 13, credit decisions, biometric identification",{"type":49,"value":952},", which is not in the captured footprint. I'm going to re-derive the applicable regimes from the company's operating jurisdictions (",{"type":44,"tag":458,"props":954,"children":955},{},[956,958],{"type":49,"value":957},"list from ",{"type":44,"tag":62,"props":959,"children":961},{"className":960},[],[962],{"type":49,"value":963},"## Company profile",{"type":49,"value":965},") and this use case's decision type (",{"type":44,"tag":458,"props":967,"children":968},{},[969],{"type":49,"value":970},"Y",{"type":49,"value":972},"), rather than use the stale footprint. If this use case is representative of work you expect to see more of, update ",{"type":44,"tag":62,"props":974,"children":976},{"className":975},[],[977],{"type":49,"value":315},{"type":49,"value":979}," at the end of this run so the next AIA doesn't have to re-derive.\"",{"type":44,"tag":131,"props":981,"children":982},{},[983],{"type":49,"value":984},"A common failure mode: the footprint lists EU AI Act + GDPR + NYC Local Law 144, and the use case is a hiring system being deployed into Illinois and Colorado. The footprint has no Illinois or Colorado entry, so iterating over it silently misses IL AIVIA, the new Colorado AI Act deployer obligations, and BIPA implications of any biometric component. Re-derive.",{"type":44,"tag":131,"props":986,"children":987},{},[988],{"type":49,"value":989},"A second failure mode: the footprint was set before a regime that now matters existed (or took effect). If re-derivation surfaces a regime not in the footprint, flag it in the output's recommendation section, cite the authority, and recommend updating the footprint.",{"type":44,"tag":131,"props":991,"children":992},{},[993,995,1000,1001,1006,1008,1013],{"type":49,"value":994},"For each regime in ",{"type":44,"tag":62,"props":996,"children":998},{"className":997},[],[999],{"type":49,"value":67},{"type":49,"value":248},{"type":44,"tag":62,"props":1002,"children":1004},{"className":1003},[],[1005],{"type":49,"value":315},{"type":49,"value":1007}," that applies to this system — ",{"type":44,"tag":135,"props":1009,"children":1010},{},[1011],{"type":49,"value":1012},"plus any regime surfaced by the re-derivation above",{"type":49,"value":1014}," — research the currently operative risk classification framework and determine where the system lands.",{"type":44,"tag":131,"props":1016,"children":1017},{},[1018],{"type":49,"value":1019},"Research tasks:",{"type":44,"tag":258,"props":1021,"children":1022},{},[1023,1028,1033,1038,1043,1048,1053],{"type":44,"tag":56,"props":1024,"children":1025},{},[1026],{"type":49,"value":1027},"What is the regime's own tier taxonomy (e.g., prohibited \u002F high-risk \u002F limited \u002F minimal, or the regime's equivalent)?",{"type":44,"tag":56,"props":1029,"children":1030},{},[1031],{"type":49,"value":1032},"What are the criteria for each tier? Cite primary sources with pinpoint references.",{"type":44,"tag":56,"props":1034,"children":1035},{},[1036],{"type":49,"value":1037},"Which tier does this system fall into given its function, affected parties, and decision consequentiality?",{"type":44,"tag":56,"props":1039,"children":1040},{},[1041],{"type":49,"value":1042},"Are there prohibited practices the system might touch? Treat any possible match as critical — flag immediately.",{"type":44,"tag":56,"props":1044,"children":1045},{},[1046],{"type":49,"value":1047},"Are there transparency obligations that apply regardless of tier (disclosure that a user is interacting with AI, labeling of AI-generated content, notice to people subject to automated decisions)?",{"type":44,"tag":56,"props":1049,"children":1050},{},[1051],{"type":49,"value":1052},"If the company is a builder providing a general-purpose or foundation model, what provider-level obligations apply (technical documentation, training data transparency, copyright compliance, systemic-risk testing)?",{"type":44,"tag":56,"props":1054,"children":1055},{},[1056,1061],{"type":44,"tag":135,"props":1057,"children":1058},{},[1059],{"type":49,"value":1060},"Does any regime in the footprint require a separate fundamental-rights impact assessment (FRIA)?",{"type":49,"value":1062}," EU AI Act Art. 27 requires a FRIA for certain deployers of high-risk AI systems (public bodies and private entities providing public services, plus certain creditworthiness and insurance-risk-assessment use cases). Check each regime for an equivalent fundamental-rights or human-rights impact assessment that is a distinct deliverable from this AIA. If a FRIA (or regime equivalent) is required, flag it as a separate deliverable in the recommendation and conditions — do not treat this AIA as a substitute.",{"type":44,"tag":131,"props":1064,"children":1065},{},[1066],{"type":49,"value":1067},"Don't assume internal-only systems are out of scope — most regimes treat employee data as personal data and employee monitoring as consequential. Verify the specific rule.",{"type":44,"tag":131,"props":1069,"children":1070},{},[1071,1084,1086,1091,1092,1097,1098,1104,1105,1111],{"type":44,"tag":135,"props":1072,"children":1073},{},[1074,1076,1082],{"type":49,"value":1075},"Provider-vs-deployer split (when ",{"type":44,"tag":62,"props":1077,"children":1079},{"className":1078},[],[1080],{"type":49,"value":1081},"AI role: Both",{"type":49,"value":1083},").",{"type":49,"value":1085}," If ",{"type":44,"tag":62,"props":1087,"children":1089},{"className":1088},[],[1090],{"type":49,"value":67},{"type":49,"value":248},{"type":44,"tag":62,"props":1093,"children":1095},{"className":1094},[],[1096],{"type":49,"value":963},{"type":49,"value":248},{"type":44,"tag":62,"props":1099,"children":1101},{"className":1100},[],[1102],{"type":49,"value":1103},"AI role",{"type":49,"value":157},{"type":44,"tag":62,"props":1106,"children":1108},{"className":1107},[],[1109],{"type":49,"value":1110},"Both",{"type":49,"value":1112}," (the company is both a provider\u002Fbuilder and a deployer), Section 6 MUST include a provider-vs-deployer mapping table per regime. Most regimes impose materially different obligations on providers (or builders) versus deployers (or users) — collapsing them into one undifferentiated list misses obligations and conflates risks. Do not combine provider and deployer obligations into a single section. Produce, per regime:",{"type":44,"tag":1114,"props":1115,"children":1116},"table",{},[1117,1141],{"type":44,"tag":1118,"props":1119,"children":1120},"thead",{},[1121],{"type":44,"tag":1122,"props":1123,"children":1124},"tr",{},[1125,1131,1136],{"type":44,"tag":1126,"props":1127,"children":1128},"th",{},[1129],{"type":49,"value":1130},"Obligation",{"type":44,"tag":1126,"props":1132,"children":1133},{},[1134],{"type":49,"value":1135},"As provider",{"type":44,"tag":1126,"props":1137,"children":1138},{},[1139],{"type":49,"value":1140},"As deployer",{"type":44,"tag":1142,"props":1143,"children":1144},"tbody",{},[1145],{"type":44,"tag":1122,"props":1146,"children":1147},{},[1148,1157,1165],{"type":44,"tag":1149,"props":1150,"children":1151},"td",{},[1152],{"type":44,"tag":458,"props":1153,"children":1154},{},[1155],{"type":49,"value":1156},"specific obligation, pinpoint cite",{"type":44,"tag":1149,"props":1158,"children":1159},{},[1160],{"type":44,"tag":458,"props":1161,"children":1162},{},[1163],{"type":49,"value":1164},"what applies \u002F does not apply \u002F with what carve-outs",{"type":44,"tag":1149,"props":1166,"children":1167},{},[1168],{"type":44,"tag":458,"props":1169,"children":1170},{},[1171],{"type":49,"value":1164},{"type":44,"tag":131,"props":1173,"children":1174},{},[1175,1180],{"type":44,"tag":135,"props":1176,"children":1177},{},[1178],{"type":49,"value":1179},"If a high-risk or equivalent classification applies:",{"type":49,"value":1181},"\nFlag in the assessment, citing the specific provision and regime. Note that this AIA documents the internal review but does not substitute for any formal conformity assessment the regime requires. Recommend external legal review before deployment in the affected jurisdiction.",{"type":44,"tag":131,"props":1183,"children":1184},{},[1185],{"type":49,"value":1186},"Capture the classification and the cited authority in the assessment output.",{"type":44,"tag":120,"props":1188,"children":1189},{},[],{"type":44,"tag":124,"props":1191,"children":1193},{"id":1192},"step-4-write-the-assessment",[1194],{"type":49,"value":1195},"Step 4: Write the assessment",{"type":44,"tag":131,"props":1197,"children":1198},{},[1199,1210],{"type":44,"tag":135,"props":1200,"children":1201},{},[1202,1204,1209],{"type":49,"value":1203},"Use the seed structure from ",{"type":44,"tag":62,"props":1205,"children":1207},{"className":1206},[],[1208],{"type":49,"value":67},{"type":49,"value":212},{"type":49,"value":1211}," If none was captured, use this default:",{"type":44,"tag":108,"props":1213,"children":1217},{"className":1214,"code":1215,"language":1216,"meta":116,"style":116},"language-markdown shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","[WORK-PRODUCT HEADER — per plugin config ## Outputs — differs by role; see `## Who's using this`]\n\n# AI Impact Assessment: [System\u002FFeature Name]\n\n**Prepared by:** [name] | **Date:** [date] | **Status:** DRAFT \u002F APPROVED\n**System owner:** [name] | **AI governance reviewer:** [name]\n**Governance tier:** [Standard \u002F Elevated \u002F High]\n**Track:** [Fast track \u002F Full assessment]\n\n---\n\n## Executive summary\n\n[Two sentences: what this AI does and whether it's okay to deploy. E.g., \"This\nsystem uses a third-party LLM to draft initial responses to customer support tickets\nbefore human agent review. Processing is consistent with the company's AI policy;\nthree conditions required before production deployment.\"]\n\n**Overall risk:** 🟢 Low \u002F 🟡 Medium \u002F 🟠 High \u002F 🔴 Very high\n\n---\n\n## 1. System description\n\n**What it does:** [plain English — not marketing]\n**Model \u002F vendor:** [who's providing the AI]\n**Deployment mode:** [Assistive \u002F Augmentative \u002F Automated]\n**Output type:** [text \u002F score \u002F classification \u002F recommendation \u002F action]\n**Status:** [Not started \u002F Pilot \u002F Production]\n\n---\n\n## 2. Affected parties\n\n**Who it acts on:** [employees \u002F customers \u002F third parties]\n**Scale:** [how many people, how often]\n**Harm if wrong:** [most realistic worst case — specific, not generic]\n**Vulnerable groups in scope:** [yes — [who] \u002F no]\n\n---\n\n## 3. Data inputs\n\n**Data categories used:** [specific fields, not \"user data\"]\n**Personal data:** [yes — [whose] \u002F no]\n**Data leaves perimeter?** [yes — to [vendor] \u002F no]\n**Model training:** [company data used \u002F foundation model \u002F fine-tuned on [dataset]]\n\n---\n\n## 4. Decision-making and oversight\n\n**Human in the loop:** [Always \u002F Nominally (rubber-stamp risk) \u002F No]\n**Override mechanism:** [how a human can intervene or correct]\n**Appeals \u002F correction for affected parties:** [yes — [how] \u002F no]\n**Named owner:** [name or role]\n\n---\n\n## 5. Accuracy and bias\n\n**Error rate:** [known \u002F estimated \u002F untested]\n**Failure mode:** [what happens when it's wrong — surfaced? logged? corrected?]\n**Bias testing:** [done — [results] \u002F not done \u002F not applicable]\n\n---\n\n## 6. Regulatory classification\n\n*[One subsection per regime in the regulatory footprint that applies to this system.]*\n\n**Regime:** [name]\n**Classification under this regime:** [tier, with pinpoint citation to the controlling provision]\n**Prohibited practices triggered:** [none identified \u002F [specific provision and why]]\n**Applicable obligations:** [researched list with citations — transparency, documentation, human oversight, testing, registration, etc.]\n**Fundamental-rights impact assessment required?** [Yes — e.g., EU AI Act Art. 27 FRIA applies \u002F regime equivalent \u002F No \u002F Not applicable. If yes, this is a separate deliverable, not subsumed by this AIA.]\n**Effective \u002F enforcement date:** [date(s)]\n**Ambiguity or open interpretation:** [flag anything not yet settled]\n\n**Provider-vs-deployer obligation split (required if `AI role: Both`):**\n\n| Obligation | As provider | As deployer |\n|---|---|---|\n| [specific obligation + pinpoint cite] | [what applies \u002F does not apply] | [what applies \u002F does not apply] |\n\n---\n\n## 7. AI policy consistency\n\n| Policy commitment | Consistent? | Notes |\n|---|---|---|\n| [commitment from `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fai-governance-legal\u002FCLAUDE.md` AI policy section] | 🟢 \u002F 🟡 \u002F 🟠 \u002F 🔴 | |\n\n[If any item is 🟡 or worse: policy update needed before deployment, or design needs to change.\nOne of them has to change — not both flagged and left open.]\n\n---\n\n## 8. Risks and mitigations\n\n| # | Risk | Likelihood | Impact | Mitigation | Status | Owner |\n|---|---|---|---|---|---|---|\n| 1 | [specific risk tied to this design — not \"AI hallucination\" generically] | L\u002FM\u002FH | L\u002FM\u002FH | [specific control] | Done \u002F Planned \u002F Gap | [name] |\n\n**Residual risk after mitigations:** [assessment]\n\n---\n\n## 9. Recommendation\n\n**[APPROVED \u002F APPROVED WITH CONDITIONS \u002F CHANGES REQUIRED \u002F NOT APPROVED]**\n\n**Conditions (if any):**\n- [ ] [specific action before deployment — owner, deadline]\n\n**Privacy review required?** [Yes — run `\u002Fprivacy-legal:pia-generation`, if the plugin is installed \u002F\nNo]\n\n**Sign-off:** [name, date]\n\n---\n\n## Cite check\n\nRegulatory citations in Section 6 (and anywhere else) were generated by an AI model and have not been verified against primary sources. Before the assessment is certified or relied on, run a verification pass against a legal research tool (Westlaw, EUR-Lex, or your firm's platform) for each cited provision — confirm the pinpoint, currency, and any delegated or implementing acts. The AI regulatory landscape shifts quickly; verify before advising. Source tags on each citation (e.g., `[EUR-Lex]`, `[web search — verify]`) show where it came from; `verify` tags carry higher fabrication risk and should be checked first.\n","markdown",[1218],{"type":44,"tag":62,"props":1219,"children":1220},{"__ignoreMap":116},[1221,1252,1262,1277,1285,1373,1431,1453,1475,1483,1492,1500,1514,1522,1531,1540,1549,1558,1566,1588,1596,1604,1612,1625,1633,1655,1677,1699,1721,1742,1750,1758,1766,1779,1787,1809,1830,1852,1893,1901,1909,1917,1930,1938,1960,1998,2037,2076,2084,2092,2100,2113,2121,2143,2165,2203,2225,2233,2241,2249,2262,2270,2292,2314,2354,2362,2370,2378,2391,2399,2420,2428,2457,2479,2501,2523,2545,2575,2597,2605,2641,2649,2686,2695,2729,2737,2745,2753,2766,2774,2809,2817,2865,2873,2882,2891,2899,2907,2915,2928,2936,3007,3016,3093,3101,3131,3139,3147,3155,3168,3176,3193,3201,3218,3232,3240,3280,3289,3297,3319,3327,3335,3343,3356,3364],{"type":44,"tag":458,"props":1222,"children":1225},{"class":1223,"line":1224},"line",1,[1226,1232,1238,1243,1247],{"type":44,"tag":458,"props":1227,"children":1229},{"style":1228},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[1230],{"type":49,"value":1231},"[WORK-PRODUCT HEADER — per plugin config ## Outputs — differs by role; see ",{"type":44,"tag":458,"props":1233,"children":1235},{"style":1234},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[1236],{"type":49,"value":1237},"`",{"type":44,"tag":458,"props":1239,"children":1241},{"style":1240},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[1242],{"type":49,"value":604},{"type":44,"tag":458,"props":1244,"children":1245},{"style":1234},[1246],{"type":49,"value":1237},{"type":44,"tag":458,"props":1248,"children":1249},{"style":1228},[1250],{"type":49,"value":1251},"]\n",{"type":44,"tag":458,"props":1253,"children":1255},{"class":1223,"line":1254},2,[1256],{"type":44,"tag":458,"props":1257,"children":1259},{"emptyLinePlaceholder":1258},true,[1260],{"type":49,"value":1261},"\n",{"type":44,"tag":458,"props":1263,"children":1265},{"class":1223,"line":1264},3,[1266,1271],{"type":44,"tag":458,"props":1267,"children":1268},{"style":1234},[1269],{"type":49,"value":1270},"# ",{"type":44,"tag":458,"props":1272,"children":1274},{"style":1273},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[1275],{"type":49,"value":1276},"AI Impact Assessment: [System\u002FFeature Name]\n",{"type":44,"tag":458,"props":1278,"children":1280},{"class":1223,"line":1279},4,[1281],{"type":44,"tag":458,"props":1282,"children":1283},{"emptyLinePlaceholder":1258},[1284],{"type":49,"value":1261},{"type":44,"tag":458,"props":1286,"children":1288},{"class":1223,"line":1287},5,[1289,1295,1301,1305,1310,1315,1320,1325,1329,1334,1338,1342,1347,1351,1355,1359,1364,1368],{"type":44,"tag":458,"props":1290,"children":1292},{"style":1291},"--shiki-light:#39ADB5;--shiki-light-font-weight:bold;--shiki-default:#89DDFF;--shiki-default-font-weight:bold;--shiki-dark:#89DDFF;--shiki-dark-font-weight:bold",[1293],{"type":49,"value":1294},"**",{"type":44,"tag":458,"props":1296,"children":1298},{"style":1297},"--shiki-light:#E53935;--shiki-light-font-weight:bold;--shiki-default:#F07178;--shiki-default-font-weight:bold;--shiki-dark:#F07178;--shiki-dark-font-weight:bold",[1299],{"type":49,"value":1300},"Prepared by:",{"type":44,"tag":458,"props":1302,"children":1303},{"style":1291},[1304],{"type":49,"value":1294},{"type":44,"tag":458,"props":1306,"children":1307},{"style":1234},[1308],{"type":49,"value":1309}," [",{"type":44,"tag":458,"props":1311,"children":1312},{"style":1240},[1313],{"type":49,"value":1314},"name",{"type":44,"tag":458,"props":1316,"children":1317},{"style":1234},[1318],{"type":49,"value":1319},"]",{"type":44,"tag":458,"props":1321,"children":1322},{"style":1228},[1323],{"type":49,"value":1324}," | ",{"type":44,"tag":458,"props":1326,"children":1327},{"style":1291},[1328],{"type":49,"value":1294},{"type":44,"tag":458,"props":1330,"children":1331},{"style":1297},[1332],{"type":49,"value":1333},"Date:",{"type":44,"tag":458,"props":1335,"children":1336},{"style":1291},[1337],{"type":49,"value":1294},{"type":44,"tag":458,"props":1339,"children":1340},{"style":1234},[1341],{"type":49,"value":1309},{"type":44,"tag":458,"props":1343,"children":1344},{"style":1240},[1345],{"type":49,"value":1346},"date",{"type":44,"tag":458,"props":1348,"children":1349},{"style":1234},[1350],{"type":49,"value":1319},{"type":44,"tag":458,"props":1352,"children":1353},{"style":1228},[1354],{"type":49,"value":1324},{"type":44,"tag":458,"props":1356,"children":1357},{"style":1291},[1358],{"type":49,"value":1294},{"type":44,"tag":458,"props":1360,"children":1361},{"style":1297},[1362],{"type":49,"value":1363},"Status:",{"type":44,"tag":458,"props":1365,"children":1366},{"style":1291},[1367],{"type":49,"value":1294},{"type":44,"tag":458,"props":1369,"children":1370},{"style":1228},[1371],{"type":49,"value":1372}," DRAFT \u002F APPROVED\n",{"type":44,"tag":458,"props":1374,"children":1376},{"class":1223,"line":1375},6,[1377,1381,1386,1390,1394,1398,1402,1406,1410,1415,1419,1423,1427],{"type":44,"tag":458,"props":1378,"children":1379},{"style":1291},[1380],{"type":49,"value":1294},{"type":44,"tag":458,"props":1382,"children":1383},{"style":1297},[1384],{"type":49,"value":1385},"System owner:",{"type":44,"tag":458,"props":1387,"children":1388},{"style":1291},[1389],{"type":49,"value":1294},{"type":44,"tag":458,"props":1391,"children":1392},{"style":1234},[1393],{"type":49,"value":1309},{"type":44,"tag":458,"props":1395,"children":1396},{"style":1240},[1397],{"type":49,"value":1314},{"type":44,"tag":458,"props":1399,"children":1400},{"style":1234},[1401],{"type":49,"value":1319},{"type":44,"tag":458,"props":1403,"children":1404},{"style":1228},[1405],{"type":49,"value":1324},{"type":44,"tag":458,"props":1407,"children":1408},{"style":1291},[1409],{"type":49,"value":1294},{"type":44,"tag":458,"props":1411,"children":1412},{"style":1297},[1413],{"type":49,"value":1414},"AI governance reviewer:",{"type":44,"tag":458,"props":1416,"children":1417},{"style":1291},[1418],{"type":49,"value":1294},{"type":44,"tag":458,"props":1420,"children":1421},{"style":1234},[1422],{"type":49,"value":1309},{"type":44,"tag":458,"props":1424,"children":1425},{"style":1240},[1426],{"type":49,"value":1314},{"type":44,"tag":458,"props":1428,"children":1429},{"style":1234},[1430],{"type":49,"value":1251},{"type":44,"tag":458,"props":1432,"children":1434},{"class":1223,"line":1433},7,[1435,1439,1444,1448],{"type":44,"tag":458,"props":1436,"children":1437},{"style":1291},[1438],{"type":49,"value":1294},{"type":44,"tag":458,"props":1440,"children":1441},{"style":1297},[1442],{"type":49,"value":1443},"Governance tier:",{"type":44,"tag":458,"props":1445,"children":1446},{"style":1291},[1447],{"type":49,"value":1294},{"type":44,"tag":458,"props":1449,"children":1450},{"style":1228},[1451],{"type":49,"value":1452}," [Standard \u002F Elevated \u002F High]\n",{"type":44,"tag":458,"props":1454,"children":1456},{"class":1223,"line":1455},8,[1457,1461,1466,1470],{"type":44,"tag":458,"props":1458,"children":1459},{"style":1291},[1460],{"type":49,"value":1294},{"type":44,"tag":458,"props":1462,"children":1463},{"style":1297},[1464],{"type":49,"value":1465},"Track:",{"type":44,"tag":458,"props":1467,"children":1468},{"style":1291},[1469],{"type":49,"value":1294},{"type":44,"tag":458,"props":1471,"children":1472},{"style":1228},[1473],{"type":49,"value":1474}," [Fast track \u002F Full assessment]\n",{"type":44,"tag":458,"props":1476,"children":1478},{"class":1223,"line":1477},9,[1479],{"type":44,"tag":458,"props":1480,"children":1481},{"emptyLinePlaceholder":1258},[1482],{"type":49,"value":1261},{"type":44,"tag":458,"props":1484,"children":1486},{"class":1223,"line":1485},10,[1487],{"type":44,"tag":458,"props":1488,"children":1489},{"style":1234},[1490],{"type":49,"value":1491},"---\n",{"type":44,"tag":458,"props":1493,"children":1495},{"class":1223,"line":1494},11,[1496],{"type":44,"tag":458,"props":1497,"children":1498},{"emptyLinePlaceholder":1258},[1499],{"type":49,"value":1261},{"type":44,"tag":458,"props":1501,"children":1503},{"class":1223,"line":1502},12,[1504,1509],{"type":44,"tag":458,"props":1505,"children":1506},{"style":1234},[1507],{"type":49,"value":1508},"## ",{"type":44,"tag":458,"props":1510,"children":1511},{"style":1273},[1512],{"type":49,"value":1513},"Executive summary\n",{"type":44,"tag":458,"props":1515,"children":1517},{"class":1223,"line":1516},13,[1518],{"type":44,"tag":458,"props":1519,"children":1520},{"emptyLinePlaceholder":1258},[1521],{"type":49,"value":1261},{"type":44,"tag":458,"props":1523,"children":1525},{"class":1223,"line":1524},14,[1526],{"type":44,"tag":458,"props":1527,"children":1528},{"style":1228},[1529],{"type":49,"value":1530},"[Two sentences: what this AI does and whether it's okay to deploy. E.g., \"This\n",{"type":44,"tag":458,"props":1532,"children":1534},{"class":1223,"line":1533},15,[1535],{"type":44,"tag":458,"props":1536,"children":1537},{"style":1228},[1538],{"type":49,"value":1539},"system uses a third-party LLM to draft initial responses to customer support tickets\n",{"type":44,"tag":458,"props":1541,"children":1543},{"class":1223,"line":1542},16,[1544],{"type":44,"tag":458,"props":1545,"children":1546},{"style":1228},[1547],{"type":49,"value":1548},"before human agent review. Processing is consistent with the company's AI policy;\n",{"type":44,"tag":458,"props":1550,"children":1552},{"class":1223,"line":1551},17,[1553],{"type":44,"tag":458,"props":1554,"children":1555},{"style":1228},[1556],{"type":49,"value":1557},"three conditions required before production deployment.\"]\n",{"type":44,"tag":458,"props":1559,"children":1561},{"class":1223,"line":1560},18,[1562],{"type":44,"tag":458,"props":1563,"children":1564},{"emptyLinePlaceholder":1258},[1565],{"type":49,"value":1261},{"type":44,"tag":458,"props":1567,"children":1569},{"class":1223,"line":1568},19,[1570,1574,1579,1583],{"type":44,"tag":458,"props":1571,"children":1572},{"style":1291},[1573],{"type":49,"value":1294},{"type":44,"tag":458,"props":1575,"children":1576},{"style":1297},[1577],{"type":49,"value":1578},"Overall risk:",{"type":44,"tag":458,"props":1580,"children":1581},{"style":1291},[1582],{"type":49,"value":1294},{"type":44,"tag":458,"props":1584,"children":1585},{"style":1228},[1586],{"type":49,"value":1587}," 🟢 Low \u002F 🟡 Medium \u002F 🟠 High \u002F 🔴 Very high\n",{"type":44,"tag":458,"props":1589,"children":1591},{"class":1223,"line":1590},20,[1592],{"type":44,"tag":458,"props":1593,"children":1594},{"emptyLinePlaceholder":1258},[1595],{"type":49,"value":1261},{"type":44,"tag":458,"props":1597,"children":1599},{"class":1223,"line":1598},21,[1600],{"type":44,"tag":458,"props":1601,"children":1602},{"style":1234},[1603],{"type":49,"value":1491},{"type":44,"tag":458,"props":1605,"children":1607},{"class":1223,"line":1606},22,[1608],{"type":44,"tag":458,"props":1609,"children":1610},{"emptyLinePlaceholder":1258},[1611],{"type":49,"value":1261},{"type":44,"tag":458,"props":1613,"children":1615},{"class":1223,"line":1614},23,[1616,1620],{"type":44,"tag":458,"props":1617,"children":1618},{"style":1234},[1619],{"type":49,"value":1508},{"type":44,"tag":458,"props":1621,"children":1622},{"style":1273},[1623],{"type":49,"value":1624},"1. System description\n",{"type":44,"tag":458,"props":1626,"children":1628},{"class":1223,"line":1627},24,[1629],{"type":44,"tag":458,"props":1630,"children":1631},{"emptyLinePlaceholder":1258},[1632],{"type":49,"value":1261},{"type":44,"tag":458,"props":1634,"children":1636},{"class":1223,"line":1635},25,[1637,1641,1646,1650],{"type":44,"tag":458,"props":1638,"children":1639},{"style":1291},[1640],{"type":49,"value":1294},{"type":44,"tag":458,"props":1642,"children":1643},{"style":1297},[1644],{"type":49,"value":1645},"What it does:",{"type":44,"tag":458,"props":1647,"children":1648},{"style":1291},[1649],{"type":49,"value":1294},{"type":44,"tag":458,"props":1651,"children":1652},{"style":1228},[1653],{"type":49,"value":1654}," [plain English — not marketing]\n",{"type":44,"tag":458,"props":1656,"children":1658},{"class":1223,"line":1657},26,[1659,1663,1668,1672],{"type":44,"tag":458,"props":1660,"children":1661},{"style":1291},[1662],{"type":49,"value":1294},{"type":44,"tag":458,"props":1664,"children":1665},{"style":1297},[1666],{"type":49,"value":1667},"Model \u002F vendor:",{"type":44,"tag":458,"props":1669,"children":1670},{"style":1291},[1671],{"type":49,"value":1294},{"type":44,"tag":458,"props":1673,"children":1674},{"style":1228},[1675],{"type":49,"value":1676}," [who's providing the AI]\n",{"type":44,"tag":458,"props":1678,"children":1680},{"class":1223,"line":1679},27,[1681,1685,1690,1694],{"type":44,"tag":458,"props":1682,"children":1683},{"style":1291},[1684],{"type":49,"value":1294},{"type":44,"tag":458,"props":1686,"children":1687},{"style":1297},[1688],{"type":49,"value":1689},"Deployment mode:",{"type":44,"tag":458,"props":1691,"children":1692},{"style":1291},[1693],{"type":49,"value":1294},{"type":44,"tag":458,"props":1695,"children":1696},{"style":1228},[1697],{"type":49,"value":1698}," [Assistive \u002F Augmentative \u002F Automated]\n",{"type":44,"tag":458,"props":1700,"children":1702},{"class":1223,"line":1701},28,[1703,1707,1712,1716],{"type":44,"tag":458,"props":1704,"children":1705},{"style":1291},[1706],{"type":49,"value":1294},{"type":44,"tag":458,"props":1708,"children":1709},{"style":1297},[1710],{"type":49,"value":1711},"Output type:",{"type":44,"tag":458,"props":1713,"children":1714},{"style":1291},[1715],{"type":49,"value":1294},{"type":44,"tag":458,"props":1717,"children":1718},{"style":1228},[1719],{"type":49,"value":1720}," [text \u002F score \u002F classification \u002F recommendation \u002F action]\n",{"type":44,"tag":458,"props":1722,"children":1724},{"class":1223,"line":1723},29,[1725,1729,1733,1737],{"type":44,"tag":458,"props":1726,"children":1727},{"style":1291},[1728],{"type":49,"value":1294},{"type":44,"tag":458,"props":1730,"children":1731},{"style":1297},[1732],{"type":49,"value":1363},{"type":44,"tag":458,"props":1734,"children":1735},{"style":1291},[1736],{"type":49,"value":1294},{"type":44,"tag":458,"props":1738,"children":1739},{"style":1228},[1740],{"type":49,"value":1741}," [Not started \u002F Pilot \u002F Production]\n",{"type":44,"tag":458,"props":1743,"children":1745},{"class":1223,"line":1744},30,[1746],{"type":44,"tag":458,"props":1747,"children":1748},{"emptyLinePlaceholder":1258},[1749],{"type":49,"value":1261},{"type":44,"tag":458,"props":1751,"children":1753},{"class":1223,"line":1752},31,[1754],{"type":44,"tag":458,"props":1755,"children":1756},{"style":1234},[1757],{"type":49,"value":1491},{"type":44,"tag":458,"props":1759,"children":1761},{"class":1223,"line":1760},32,[1762],{"type":44,"tag":458,"props":1763,"children":1764},{"emptyLinePlaceholder":1258},[1765],{"type":49,"value":1261},{"type":44,"tag":458,"props":1767,"children":1769},{"class":1223,"line":1768},33,[1770,1774],{"type":44,"tag":458,"props":1771,"children":1772},{"style":1234},[1773],{"type":49,"value":1508},{"type":44,"tag":458,"props":1775,"children":1776},{"style":1273},[1777],{"type":49,"value":1778},"2. Affected parties\n",{"type":44,"tag":458,"props":1780,"children":1782},{"class":1223,"line":1781},34,[1783],{"type":44,"tag":458,"props":1784,"children":1785},{"emptyLinePlaceholder":1258},[1786],{"type":49,"value":1261},{"type":44,"tag":458,"props":1788,"children":1790},{"class":1223,"line":1789},35,[1791,1795,1800,1804],{"type":44,"tag":458,"props":1792,"children":1793},{"style":1291},[1794],{"type":49,"value":1294},{"type":44,"tag":458,"props":1796,"children":1797},{"style":1297},[1798],{"type":49,"value":1799},"Who it acts on:",{"type":44,"tag":458,"props":1801,"children":1802},{"style":1291},[1803],{"type":49,"value":1294},{"type":44,"tag":458,"props":1805,"children":1806},{"style":1228},[1807],{"type":49,"value":1808}," [employees \u002F customers \u002F third parties]\n",{"type":44,"tag":458,"props":1810,"children":1812},{"class":1223,"line":1811},36,[1813,1817,1821,1825],{"type":44,"tag":458,"props":1814,"children":1815},{"style":1291},[1816],{"type":49,"value":1294},{"type":44,"tag":458,"props":1818,"children":1819},{"style":1297},[1820],{"type":49,"value":867},{"type":44,"tag":458,"props":1822,"children":1823},{"style":1291},[1824],{"type":49,"value":1294},{"type":44,"tag":458,"props":1826,"children":1827},{"style":1228},[1828],{"type":49,"value":1829}," [how many people, how often]\n",{"type":44,"tag":458,"props":1831,"children":1833},{"class":1223,"line":1832},37,[1834,1838,1843,1847],{"type":44,"tag":458,"props":1835,"children":1836},{"style":1291},[1837],{"type":49,"value":1294},{"type":44,"tag":458,"props":1839,"children":1840},{"style":1297},[1841],{"type":49,"value":1842},"Harm if wrong:",{"type":44,"tag":458,"props":1844,"children":1845},{"style":1291},[1846],{"type":49,"value":1294},{"type":44,"tag":458,"props":1848,"children":1849},{"style":1228},[1850],{"type":49,"value":1851}," [most realistic worst case — specific, not generic]\n",{"type":44,"tag":458,"props":1853,"children":1855},{"class":1223,"line":1854},38,[1856,1860,1865,1869,1874,1879,1884,1888],{"type":44,"tag":458,"props":1857,"children":1858},{"style":1291},[1859],{"type":49,"value":1294},{"type":44,"tag":458,"props":1861,"children":1862},{"style":1297},[1863],{"type":49,"value":1864},"Vulnerable groups in scope:",{"type":44,"tag":458,"props":1866,"children":1867},{"style":1291},[1868],{"type":49,"value":1294},{"type":44,"tag":458,"props":1870,"children":1871},{"style":1228},[1872],{"type":49,"value":1873}," [yes — ",{"type":44,"tag":458,"props":1875,"children":1876},{"style":1234},[1877],{"type":49,"value":1878},"[",{"type":44,"tag":458,"props":1880,"children":1881},{"style":1240},[1882],{"type":49,"value":1883},"who",{"type":44,"tag":458,"props":1885,"children":1886},{"style":1234},[1887],{"type":49,"value":1319},{"type":44,"tag":458,"props":1889,"children":1890},{"style":1228},[1891],{"type":49,"value":1892}," \u002F no]\n",{"type":44,"tag":458,"props":1894,"children":1896},{"class":1223,"line":1895},39,[1897],{"type":44,"tag":458,"props":1898,"children":1899},{"emptyLinePlaceholder":1258},[1900],{"type":49,"value":1261},{"type":44,"tag":458,"props":1902,"children":1904},{"class":1223,"line":1903},40,[1905],{"type":44,"tag":458,"props":1906,"children":1907},{"style":1234},[1908],{"type":49,"value":1491},{"type":44,"tag":458,"props":1910,"children":1912},{"class":1223,"line":1911},41,[1913],{"type":44,"tag":458,"props":1914,"children":1915},{"emptyLinePlaceholder":1258},[1916],{"type":49,"value":1261},{"type":44,"tag":458,"props":1918,"children":1920},{"class":1223,"line":1919},42,[1921,1925],{"type":44,"tag":458,"props":1922,"children":1923},{"style":1234},[1924],{"type":49,"value":1508},{"type":44,"tag":458,"props":1926,"children":1927},{"style":1273},[1928],{"type":49,"value":1929},"3. Data inputs\n",{"type":44,"tag":458,"props":1931,"children":1933},{"class":1223,"line":1932},43,[1934],{"type":44,"tag":458,"props":1935,"children":1936},{"emptyLinePlaceholder":1258},[1937],{"type":49,"value":1261},{"type":44,"tag":458,"props":1939,"children":1941},{"class":1223,"line":1940},44,[1942,1946,1951,1955],{"type":44,"tag":458,"props":1943,"children":1944},{"style":1291},[1945],{"type":49,"value":1294},{"type":44,"tag":458,"props":1947,"children":1948},{"style":1297},[1949],{"type":49,"value":1950},"Data categories used:",{"type":44,"tag":458,"props":1952,"children":1953},{"style":1291},[1954],{"type":49,"value":1294},{"type":44,"tag":458,"props":1956,"children":1957},{"style":1228},[1958],{"type":49,"value":1959}," [specific fields, not \"user data\"]\n",{"type":44,"tag":458,"props":1961,"children":1963},{"class":1223,"line":1962},45,[1964,1968,1973,1977,1981,1985,1990,1994],{"type":44,"tag":458,"props":1965,"children":1966},{"style":1291},[1967],{"type":49,"value":1294},{"type":44,"tag":458,"props":1969,"children":1970},{"style":1297},[1971],{"type":49,"value":1972},"Personal data:",{"type":44,"tag":458,"props":1974,"children":1975},{"style":1291},[1976],{"type":49,"value":1294},{"type":44,"tag":458,"props":1978,"children":1979},{"style":1228},[1980],{"type":49,"value":1873},{"type":44,"tag":458,"props":1982,"children":1983},{"style":1234},[1984],{"type":49,"value":1878},{"type":44,"tag":458,"props":1986,"children":1987},{"style":1240},[1988],{"type":49,"value":1989},"whose",{"type":44,"tag":458,"props":1991,"children":1992},{"style":1234},[1993],{"type":49,"value":1319},{"type":44,"tag":458,"props":1995,"children":1996},{"style":1228},[1997],{"type":49,"value":1892},{"type":44,"tag":458,"props":1999,"children":2001},{"class":1223,"line":2000},46,[2002,2006,2011,2015,2020,2024,2029,2033],{"type":44,"tag":458,"props":2003,"children":2004},{"style":1291},[2005],{"type":49,"value":1294},{"type":44,"tag":458,"props":2007,"children":2008},{"style":1297},[2009],{"type":49,"value":2010},"Data leaves perimeter?",{"type":44,"tag":458,"props":2012,"children":2013},{"style":1291},[2014],{"type":49,"value":1294},{"type":44,"tag":458,"props":2016,"children":2017},{"style":1228},[2018],{"type":49,"value":2019}," [yes — to ",{"type":44,"tag":458,"props":2021,"children":2022},{"style":1234},[2023],{"type":49,"value":1878},{"type":44,"tag":458,"props":2025,"children":2026},{"style":1240},[2027],{"type":49,"value":2028},"vendor",{"type":44,"tag":458,"props":2030,"children":2031},{"style":1234},[2032],{"type":49,"value":1319},{"type":44,"tag":458,"props":2034,"children":2035},{"style":1228},[2036],{"type":49,"value":1892},{"type":44,"tag":458,"props":2038,"children":2040},{"class":1223,"line":2039},47,[2041,2045,2050,2054,2059,2063,2068,2072],{"type":44,"tag":458,"props":2042,"children":2043},{"style":1291},[2044],{"type":49,"value":1294},{"type":44,"tag":458,"props":2046,"children":2047},{"style":1297},[2048],{"type":49,"value":2049},"Model training:",{"type":44,"tag":458,"props":2051,"children":2052},{"style":1291},[2053],{"type":49,"value":1294},{"type":44,"tag":458,"props":2055,"children":2056},{"style":1228},[2057],{"type":49,"value":2058}," [company data used \u002F foundation model \u002F fine-tuned on ",{"type":44,"tag":458,"props":2060,"children":2061},{"style":1234},[2062],{"type":49,"value":1878},{"type":44,"tag":458,"props":2064,"children":2065},{"style":1240},[2066],{"type":49,"value":2067},"dataset",{"type":44,"tag":458,"props":2069,"children":2070},{"style":1234},[2071],{"type":49,"value":1319},{"type":44,"tag":458,"props":2073,"children":2074},{"style":1228},[2075],{"type":49,"value":1251},{"type":44,"tag":458,"props":2077,"children":2079},{"class":1223,"line":2078},48,[2080],{"type":44,"tag":458,"props":2081,"children":2082},{"emptyLinePlaceholder":1258},[2083],{"type":49,"value":1261},{"type":44,"tag":458,"props":2085,"children":2087},{"class":1223,"line":2086},49,[2088],{"type":44,"tag":458,"props":2089,"children":2090},{"style":1234},[2091],{"type":49,"value":1491},{"type":44,"tag":458,"props":2093,"children":2095},{"class":1223,"line":2094},50,[2096],{"type":44,"tag":458,"props":2097,"children":2098},{"emptyLinePlaceholder":1258},[2099],{"type":49,"value":1261},{"type":44,"tag":458,"props":2101,"children":2103},{"class":1223,"line":2102},51,[2104,2108],{"type":44,"tag":458,"props":2105,"children":2106},{"style":1234},[2107],{"type":49,"value":1508},{"type":44,"tag":458,"props":2109,"children":2110},{"style":1273},[2111],{"type":49,"value":2112},"4. Decision-making and oversight\n",{"type":44,"tag":458,"props":2114,"children":2116},{"class":1223,"line":2115},52,[2117],{"type":44,"tag":458,"props":2118,"children":2119},{"emptyLinePlaceholder":1258},[2120],{"type":49,"value":1261},{"type":44,"tag":458,"props":2122,"children":2124},{"class":1223,"line":2123},53,[2125,2129,2134,2138],{"type":44,"tag":458,"props":2126,"children":2127},{"style":1291},[2128],{"type":49,"value":1294},{"type":44,"tag":458,"props":2130,"children":2131},{"style":1297},[2132],{"type":49,"value":2133},"Human in the loop:",{"type":44,"tag":458,"props":2135,"children":2136},{"style":1291},[2137],{"type":49,"value":1294},{"type":44,"tag":458,"props":2139,"children":2140},{"style":1228},[2141],{"type":49,"value":2142}," [Always \u002F Nominally (rubber-stamp risk) \u002F No]\n",{"type":44,"tag":458,"props":2144,"children":2146},{"class":1223,"line":2145},54,[2147,2151,2156,2160],{"type":44,"tag":458,"props":2148,"children":2149},{"style":1291},[2150],{"type":49,"value":1294},{"type":44,"tag":458,"props":2152,"children":2153},{"style":1297},[2154],{"type":49,"value":2155},"Override mechanism:",{"type":44,"tag":458,"props":2157,"children":2158},{"style":1291},[2159],{"type":49,"value":1294},{"type":44,"tag":458,"props":2161,"children":2162},{"style":1228},[2163],{"type":49,"value":2164}," [how a human can intervene or correct]\n",{"type":44,"tag":458,"props":2166,"children":2168},{"class":1223,"line":2167},55,[2169,2173,2178,2182,2186,2190,2195,2199],{"type":44,"tag":458,"props":2170,"children":2171},{"style":1291},[2172],{"type":49,"value":1294},{"type":44,"tag":458,"props":2174,"children":2175},{"style":1297},[2176],{"type":49,"value":2177},"Appeals \u002F correction for affected parties:",{"type":44,"tag":458,"props":2179,"children":2180},{"style":1291},[2181],{"type":49,"value":1294},{"type":44,"tag":458,"props":2183,"children":2184},{"style":1228},[2185],{"type":49,"value":1873},{"type":44,"tag":458,"props":2187,"children":2188},{"style":1234},[2189],{"type":49,"value":1878},{"type":44,"tag":458,"props":2191,"children":2192},{"style":1240},[2193],{"type":49,"value":2194},"how",{"type":44,"tag":458,"props":2196,"children":2197},{"style":1234},[2198],{"type":49,"value":1319},{"type":44,"tag":458,"props":2200,"children":2201},{"style":1228},[2202],{"type":49,"value":1892},{"type":44,"tag":458,"props":2204,"children":2206},{"class":1223,"line":2205},56,[2207,2211,2216,2220],{"type":44,"tag":458,"props":2208,"children":2209},{"style":1291},[2210],{"type":49,"value":1294},{"type":44,"tag":458,"props":2212,"children":2213},{"style":1297},[2214],{"type":49,"value":2215},"Named owner:",{"type":44,"tag":458,"props":2217,"children":2218},{"style":1291},[2219],{"type":49,"value":1294},{"type":44,"tag":458,"props":2221,"children":2222},{"style":1228},[2223],{"type":49,"value":2224}," [name or role]\n",{"type":44,"tag":458,"props":2226,"children":2228},{"class":1223,"line":2227},57,[2229],{"type":44,"tag":458,"props":2230,"children":2231},{"emptyLinePlaceholder":1258},[2232],{"type":49,"value":1261},{"type":44,"tag":458,"props":2234,"children":2236},{"class":1223,"line":2235},58,[2237],{"type":44,"tag":458,"props":2238,"children":2239},{"style":1234},[2240],{"type":49,"value":1491},{"type":44,"tag":458,"props":2242,"children":2244},{"class":1223,"line":2243},59,[2245],{"type":44,"tag":458,"props":2246,"children":2247},{"emptyLinePlaceholder":1258},[2248],{"type":49,"value":1261},{"type":44,"tag":458,"props":2250,"children":2252},{"class":1223,"line":2251},60,[2253,2257],{"type":44,"tag":458,"props":2254,"children":2255},{"style":1234},[2256],{"type":49,"value":1508},{"type":44,"tag":458,"props":2258,"children":2259},{"style":1273},[2260],{"type":49,"value":2261},"5. Accuracy and bias\n",{"type":44,"tag":458,"props":2263,"children":2265},{"class":1223,"line":2264},61,[2266],{"type":44,"tag":458,"props":2267,"children":2268},{"emptyLinePlaceholder":1258},[2269],{"type":49,"value":1261},{"type":44,"tag":458,"props":2271,"children":2273},{"class":1223,"line":2272},62,[2274,2278,2283,2287],{"type":44,"tag":458,"props":2275,"children":2276},{"style":1291},[2277],{"type":49,"value":1294},{"type":44,"tag":458,"props":2279,"children":2280},{"style":1297},[2281],{"type":49,"value":2282},"Error rate:",{"type":44,"tag":458,"props":2284,"children":2285},{"style":1291},[2286],{"type":49,"value":1294},{"type":44,"tag":458,"props":2288,"children":2289},{"style":1228},[2290],{"type":49,"value":2291}," [known \u002F estimated \u002F untested]\n",{"type":44,"tag":458,"props":2293,"children":2295},{"class":1223,"line":2294},63,[2296,2300,2305,2309],{"type":44,"tag":458,"props":2297,"children":2298},{"style":1291},[2299],{"type":49,"value":1294},{"type":44,"tag":458,"props":2301,"children":2302},{"style":1297},[2303],{"type":49,"value":2304},"Failure mode:",{"type":44,"tag":458,"props":2306,"children":2307},{"style":1291},[2308],{"type":49,"value":1294},{"type":44,"tag":458,"props":2310,"children":2311},{"style":1228},[2312],{"type":49,"value":2313}," [what happens when it's wrong — surfaced? logged? corrected?]\n",{"type":44,"tag":458,"props":2315,"children":2317},{"class":1223,"line":2316},64,[2318,2322,2327,2331,2336,2340,2345,2349],{"type":44,"tag":458,"props":2319,"children":2320},{"style":1291},[2321],{"type":49,"value":1294},{"type":44,"tag":458,"props":2323,"children":2324},{"style":1297},[2325],{"type":49,"value":2326},"Bias testing:",{"type":44,"tag":458,"props":2328,"children":2329},{"style":1291},[2330],{"type":49,"value":1294},{"type":44,"tag":458,"props":2332,"children":2333},{"style":1228},[2334],{"type":49,"value":2335}," [done — ",{"type":44,"tag":458,"props":2337,"children":2338},{"style":1234},[2339],{"type":49,"value":1878},{"type":44,"tag":458,"props":2341,"children":2342},{"style":1240},[2343],{"type":49,"value":2344},"results",{"type":44,"tag":458,"props":2346,"children":2347},{"style":1234},[2348],{"type":49,"value":1319},{"type":44,"tag":458,"props":2350,"children":2351},{"style":1228},[2352],{"type":49,"value":2353}," \u002F not done \u002F not applicable]\n",{"type":44,"tag":458,"props":2355,"children":2357},{"class":1223,"line":2356},65,[2358],{"type":44,"tag":458,"props":2359,"children":2360},{"emptyLinePlaceholder":1258},[2361],{"type":49,"value":1261},{"type":44,"tag":458,"props":2363,"children":2365},{"class":1223,"line":2364},66,[2366],{"type":44,"tag":458,"props":2367,"children":2368},{"style":1234},[2369],{"type":49,"value":1491},{"type":44,"tag":458,"props":2371,"children":2373},{"class":1223,"line":2372},67,[2374],{"type":44,"tag":458,"props":2375,"children":2376},{"emptyLinePlaceholder":1258},[2377],{"type":49,"value":1261},{"type":44,"tag":458,"props":2379,"children":2381},{"class":1223,"line":2380},68,[2382,2386],{"type":44,"tag":458,"props":2383,"children":2384},{"style":1234},[2385],{"type":49,"value":1508},{"type":44,"tag":458,"props":2387,"children":2388},{"style":1273},[2389],{"type":49,"value":2390},"6. Regulatory classification\n",{"type":44,"tag":458,"props":2392,"children":2394},{"class":1223,"line":2393},69,[2395],{"type":44,"tag":458,"props":2396,"children":2397},{"emptyLinePlaceholder":1258},[2398],{"type":49,"value":1261},{"type":44,"tag":458,"props":2400,"children":2402},{"class":1223,"line":2401},70,[2403,2409,2415],{"type":44,"tag":458,"props":2404,"children":2406},{"style":2405},"--shiki-light:#39ADB5;--shiki-light-font-style:italic;--shiki-default:#89DDFF;--shiki-default-font-style:italic;--shiki-dark:#89DDFF;--shiki-dark-font-style:italic",[2407],{"type":49,"value":2408},"*",{"type":44,"tag":458,"props":2410,"children":2412},{"style":2411},"--shiki-light:#E53935;--shiki-light-font-style:italic;--shiki-default:#F07178;--shiki-default-font-style:italic;--shiki-dark:#F07178;--shiki-dark-font-style:italic",[2413],{"type":49,"value":2414},"[One subsection per regime in the regulatory footprint that applies to this system.]",{"type":44,"tag":458,"props":2416,"children":2417},{"style":2405},[2418],{"type":49,"value":2419},"*\n",{"type":44,"tag":458,"props":2421,"children":2423},{"class":1223,"line":2422},71,[2424],{"type":44,"tag":458,"props":2425,"children":2426},{"emptyLinePlaceholder":1258},[2427],{"type":49,"value":1261},{"type":44,"tag":458,"props":2429,"children":2431},{"class":1223,"line":2430},72,[2432,2436,2441,2445,2449,2453],{"type":44,"tag":458,"props":2433,"children":2434},{"style":1291},[2435],{"type":49,"value":1294},{"type":44,"tag":458,"props":2437,"children":2438},{"style":1297},[2439],{"type":49,"value":2440},"Regime:",{"type":44,"tag":458,"props":2442,"children":2443},{"style":1291},[2444],{"type":49,"value":1294},{"type":44,"tag":458,"props":2446,"children":2447},{"style":1234},[2448],{"type":49,"value":1309},{"type":44,"tag":458,"props":2450,"children":2451},{"style":1240},[2452],{"type":49,"value":1314},{"type":44,"tag":458,"props":2454,"children":2455},{"style":1234},[2456],{"type":49,"value":1251},{"type":44,"tag":458,"props":2458,"children":2460},{"class":1223,"line":2459},73,[2461,2465,2470,2474],{"type":44,"tag":458,"props":2462,"children":2463},{"style":1291},[2464],{"type":49,"value":1294},{"type":44,"tag":458,"props":2466,"children":2467},{"style":1297},[2468],{"type":49,"value":2469},"Classification under this regime:",{"type":44,"tag":458,"props":2471,"children":2472},{"style":1291},[2473],{"type":49,"value":1294},{"type":44,"tag":458,"props":2475,"children":2476},{"style":1228},[2477],{"type":49,"value":2478}," [tier, with pinpoint citation to the controlling provision]\n",{"type":44,"tag":458,"props":2480,"children":2482},{"class":1223,"line":2481},74,[2483,2487,2492,2496],{"type":44,"tag":458,"props":2484,"children":2485},{"style":1291},[2486],{"type":49,"value":1294},{"type":44,"tag":458,"props":2488,"children":2489},{"style":1297},[2490],{"type":49,"value":2491},"Prohibited practices triggered:",{"type":44,"tag":458,"props":2493,"children":2494},{"style":1291},[2495],{"type":49,"value":1294},{"type":44,"tag":458,"props":2497,"children":2498},{"style":1228},[2499],{"type":49,"value":2500}," [none identified \u002F [specific provision and why]]\n",{"type":44,"tag":458,"props":2502,"children":2504},{"class":1223,"line":2503},75,[2505,2509,2514,2518],{"type":44,"tag":458,"props":2506,"children":2507},{"style":1291},[2508],{"type":49,"value":1294},{"type":44,"tag":458,"props":2510,"children":2511},{"style":1297},[2512],{"type":49,"value":2513},"Applicable obligations:",{"type":44,"tag":458,"props":2515,"children":2516},{"style":1291},[2517],{"type":49,"value":1294},{"type":44,"tag":458,"props":2519,"children":2520},{"style":1228},[2521],{"type":49,"value":2522}," [researched list with citations — transparency, documentation, human oversight, testing, registration, etc.]\n",{"type":44,"tag":458,"props":2524,"children":2526},{"class":1223,"line":2525},76,[2527,2531,2536,2540],{"type":44,"tag":458,"props":2528,"children":2529},{"style":1291},[2530],{"type":49,"value":1294},{"type":44,"tag":458,"props":2532,"children":2533},{"style":1297},[2534],{"type":49,"value":2535},"Fundamental-rights impact assessment required?",{"type":44,"tag":458,"props":2537,"children":2538},{"style":1291},[2539],{"type":49,"value":1294},{"type":44,"tag":458,"props":2541,"children":2542},{"style":1228},[2543],{"type":49,"value":2544}," [Yes — e.g., EU AI Act Art. 27 FRIA applies \u002F regime equivalent \u002F No \u002F Not applicable. If yes, this is a separate deliverable, not subsumed by this AIA.]\n",{"type":44,"tag":458,"props":2546,"children":2548},{"class":1223,"line":2547},77,[2549,2553,2558,2562,2566,2571],{"type":44,"tag":458,"props":2550,"children":2551},{"style":1291},[2552],{"type":49,"value":1294},{"type":44,"tag":458,"props":2554,"children":2555},{"style":1297},[2556],{"type":49,"value":2557},"Effective \u002F enforcement date:",{"type":44,"tag":458,"props":2559,"children":2560},{"style":1291},[2561],{"type":49,"value":1294},{"type":44,"tag":458,"props":2563,"children":2564},{"style":1234},[2565],{"type":49,"value":1309},{"type":44,"tag":458,"props":2567,"children":2568},{"style":1240},[2569],{"type":49,"value":2570},"date(s)",{"type":44,"tag":458,"props":2572,"children":2573},{"style":1234},[2574],{"type":49,"value":1251},{"type":44,"tag":458,"props":2576,"children":2578},{"class":1223,"line":2577},78,[2579,2583,2588,2592],{"type":44,"tag":458,"props":2580,"children":2581},{"style":1291},[2582],{"type":49,"value":1294},{"type":44,"tag":458,"props":2584,"children":2585},{"style":1297},[2586],{"type":49,"value":2587},"Ambiguity or open interpretation:",{"type":44,"tag":458,"props":2589,"children":2590},{"style":1291},[2591],{"type":49,"value":1294},{"type":44,"tag":458,"props":2593,"children":2594},{"style":1228},[2595],{"type":49,"value":2596}," [flag anything not yet settled]\n",{"type":44,"tag":458,"props":2598,"children":2600},{"class":1223,"line":2599},79,[2601],{"type":44,"tag":458,"props":2602,"children":2603},{"emptyLinePlaceholder":1258},[2604],{"type":49,"value":1261},{"type":44,"tag":458,"props":2606,"children":2608},{"class":1223,"line":2607},80,[2609,2613,2618,2622,2627,2631,2636],{"type":44,"tag":458,"props":2610,"children":2611},{"style":1291},[2612],{"type":49,"value":1294},{"type":44,"tag":458,"props":2614,"children":2615},{"style":1297},[2616],{"type":49,"value":2617},"Provider-vs-deployer obligation split (required if ",{"type":44,"tag":458,"props":2619,"children":2620},{"style":1291},[2621],{"type":49,"value":1237},{"type":44,"tag":458,"props":2623,"children":2625},{"style":2624},"--shiki-light:#91B859;--shiki-light-font-weight:bold;--shiki-default:#C3E88D;--shiki-default-font-weight:bold;--shiki-dark:#C3E88D;--shiki-dark-font-weight:bold",[2626],{"type":49,"value":1081},{"type":44,"tag":458,"props":2628,"children":2629},{"style":1291},[2630],{"type":49,"value":1237},{"type":44,"tag":458,"props":2632,"children":2633},{"style":1297},[2634],{"type":49,"value":2635},"):",{"type":44,"tag":458,"props":2637,"children":2638},{"style":1291},[2639],{"type":49,"value":2640},"**\n",{"type":44,"tag":458,"props":2642,"children":2644},{"class":1223,"line":2643},81,[2645],{"type":44,"tag":458,"props":2646,"children":2647},{"emptyLinePlaceholder":1258},[2648],{"type":49,"value":1261},{"type":44,"tag":458,"props":2650,"children":2652},{"class":1223,"line":2651},82,[2653,2658,2663,2667,2672,2676,2681],{"type":44,"tag":458,"props":2654,"children":2655},{"style":1234},[2656],{"type":49,"value":2657},"|",{"type":44,"tag":458,"props":2659,"children":2660},{"style":1228},[2661],{"type":49,"value":2662}," Obligation ",{"type":44,"tag":458,"props":2664,"children":2665},{"style":1234},[2666],{"type":49,"value":2657},{"type":44,"tag":458,"props":2668,"children":2669},{"style":1228},[2670],{"type":49,"value":2671}," As provider ",{"type":44,"tag":458,"props":2673,"children":2674},{"style":1234},[2675],{"type":49,"value":2657},{"type":44,"tag":458,"props":2677,"children":2678},{"style":1228},[2679],{"type":49,"value":2680}," As deployer ",{"type":44,"tag":458,"props":2682,"children":2683},{"style":1234},[2684],{"type":49,"value":2685},"|\n",{"type":44,"tag":458,"props":2687,"children":2689},{"class":1223,"line":2688},83,[2690],{"type":44,"tag":458,"props":2691,"children":2692},{"style":1234},[2693],{"type":49,"value":2694},"|---|---|---|\n",{"type":44,"tag":458,"props":2696,"children":2698},{"class":1223,"line":2697},84,[2699,2703,2708,2712,2717,2721,2725],{"type":44,"tag":458,"props":2700,"children":2701},{"style":1234},[2702],{"type":49,"value":2657},{"type":44,"tag":458,"props":2704,"children":2705},{"style":1228},[2706],{"type":49,"value":2707}," [specific obligation + pinpoint cite] ",{"type":44,"tag":458,"props":2709,"children":2710},{"style":1234},[2711],{"type":49,"value":2657},{"type":44,"tag":458,"props":2713,"children":2714},{"style":1228},[2715],{"type":49,"value":2716}," [what applies \u002F does not apply] ",{"type":44,"tag":458,"props":2718,"children":2719},{"style":1234},[2720],{"type":49,"value":2657},{"type":44,"tag":458,"props":2722,"children":2723},{"style":1228},[2724],{"type":49,"value":2716},{"type":44,"tag":458,"props":2726,"children":2727},{"style":1234},[2728],{"type":49,"value":2685},{"type":44,"tag":458,"props":2730,"children":2732},{"class":1223,"line":2731},85,[2733],{"type":44,"tag":458,"props":2734,"children":2735},{"emptyLinePlaceholder":1258},[2736],{"type":49,"value":1261},{"type":44,"tag":458,"props":2738,"children":2740},{"class":1223,"line":2739},86,[2741],{"type":44,"tag":458,"props":2742,"children":2743},{"style":1228},[2744],{"type":49,"value":1491},{"type":44,"tag":458,"props":2746,"children":2748},{"class":1223,"line":2747},87,[2749],{"type":44,"tag":458,"props":2750,"children":2751},{"emptyLinePlaceholder":1258},[2752],{"type":49,"value":1261},{"type":44,"tag":458,"props":2754,"children":2756},{"class":1223,"line":2755},88,[2757,2761],{"type":44,"tag":458,"props":2758,"children":2759},{"style":1234},[2760],{"type":49,"value":1508},{"type":44,"tag":458,"props":2762,"children":2763},{"style":1273},[2764],{"type":49,"value":2765},"7. AI policy consistency\n",{"type":44,"tag":458,"props":2767,"children":2769},{"class":1223,"line":2768},89,[2770],{"type":44,"tag":458,"props":2771,"children":2772},{"emptyLinePlaceholder":1258},[2773],{"type":49,"value":1261},{"type":44,"tag":458,"props":2775,"children":2777},{"class":1223,"line":2776},90,[2778,2782,2787,2791,2796,2800,2805],{"type":44,"tag":458,"props":2779,"children":2780},{"style":1234},[2781],{"type":49,"value":2657},{"type":44,"tag":458,"props":2783,"children":2784},{"style":1228},[2785],{"type":49,"value":2786}," Policy commitment ",{"type":44,"tag":458,"props":2788,"children":2789},{"style":1234},[2790],{"type":49,"value":2657},{"type":44,"tag":458,"props":2792,"children":2793},{"style":1228},[2794],{"type":49,"value":2795}," Consistent? ",{"type":44,"tag":458,"props":2797,"children":2798},{"style":1234},[2799],{"type":49,"value":2657},{"type":44,"tag":458,"props":2801,"children":2802},{"style":1228},[2803],{"type":49,"value":2804}," Notes ",{"type":44,"tag":458,"props":2806,"children":2807},{"style":1234},[2808],{"type":49,"value":2685},{"type":44,"tag":458,"props":2810,"children":2812},{"class":1223,"line":2811},91,[2813],{"type":44,"tag":458,"props":2814,"children":2815},{"style":1234},[2816],{"type":49,"value":2694},{"type":44,"tag":458,"props":2818,"children":2820},{"class":1223,"line":2819},92,[2821,2825,2830,2834,2838,2842,2847,2851,2856,2860],{"type":44,"tag":458,"props":2822,"children":2823},{"style":1234},[2824],{"type":49,"value":2657},{"type":44,"tag":458,"props":2826,"children":2827},{"style":1228},[2828],{"type":49,"value":2829}," [commitment from ",{"type":44,"tag":458,"props":2831,"children":2832},{"style":1234},[2833],{"type":49,"value":1237},{"type":44,"tag":458,"props":2835,"children":2836},{"style":1240},[2837],{"type":49,"value":67},{"type":44,"tag":458,"props":2839,"children":2840},{"style":1234},[2841],{"type":49,"value":1237},{"type":44,"tag":458,"props":2843,"children":2844},{"style":1228},[2845],{"type":49,"value":2846}," AI policy section] ",{"type":44,"tag":458,"props":2848,"children":2849},{"style":1234},[2850],{"type":49,"value":2657},{"type":44,"tag":458,"props":2852,"children":2853},{"style":1228},[2854],{"type":49,"value":2855}," 🟢 \u002F 🟡 \u002F 🟠 \u002F 🔴 ",{"type":44,"tag":458,"props":2857,"children":2858},{"style":1234},[2859],{"type":49,"value":2657},{"type":44,"tag":458,"props":2861,"children":2862},{"style":1234},[2863],{"type":49,"value":2864}," |\n",{"type":44,"tag":458,"props":2866,"children":2868},{"class":1223,"line":2867},93,[2869],{"type":44,"tag":458,"props":2870,"children":2871},{"emptyLinePlaceholder":1258},[2872],{"type":49,"value":1261},{"type":44,"tag":458,"props":2874,"children":2876},{"class":1223,"line":2875},94,[2877],{"type":44,"tag":458,"props":2878,"children":2879},{"style":1228},[2880],{"type":49,"value":2881},"[If any item is 🟡 or worse: policy update needed before deployment, or design needs to change.\n",{"type":44,"tag":458,"props":2883,"children":2885},{"class":1223,"line":2884},95,[2886],{"type":44,"tag":458,"props":2887,"children":2888},{"style":1228},[2889],{"type":49,"value":2890},"One of them has to change — not both flagged and left open.]\n",{"type":44,"tag":458,"props":2892,"children":2894},{"class":1223,"line":2893},96,[2895],{"type":44,"tag":458,"props":2896,"children":2897},{"emptyLinePlaceholder":1258},[2898],{"type":49,"value":1261},{"type":44,"tag":458,"props":2900,"children":2902},{"class":1223,"line":2901},97,[2903],{"type":44,"tag":458,"props":2904,"children":2905},{"style":1234},[2906],{"type":49,"value":1491},{"type":44,"tag":458,"props":2908,"children":2910},{"class":1223,"line":2909},98,[2911],{"type":44,"tag":458,"props":2912,"children":2913},{"emptyLinePlaceholder":1258},[2914],{"type":49,"value":1261},{"type":44,"tag":458,"props":2916,"children":2918},{"class":1223,"line":2917},99,[2919,2923],{"type":44,"tag":458,"props":2920,"children":2921},{"style":1234},[2922],{"type":49,"value":1508},{"type":44,"tag":458,"props":2924,"children":2925},{"style":1273},[2926],{"type":49,"value":2927},"8. Risks and mitigations\n",{"type":44,"tag":458,"props":2929,"children":2931},{"class":1223,"line":2930},100,[2932],{"type":44,"tag":458,"props":2933,"children":2934},{"emptyLinePlaceholder":1258},[2935],{"type":49,"value":1261},{"type":44,"tag":458,"props":2937,"children":2939},{"class":1223,"line":2938},101,[2940,2944,2949,2953,2958,2962,2967,2971,2976,2980,2985,2989,2994,2998,3003],{"type":44,"tag":458,"props":2941,"children":2942},{"style":1234},[2943],{"type":49,"value":2657},{"type":44,"tag":458,"props":2945,"children":2946},{"style":1228},[2947],{"type":49,"value":2948}," # ",{"type":44,"tag":458,"props":2950,"children":2951},{"style":1234},[2952],{"type":49,"value":2657},{"type":44,"tag":458,"props":2954,"children":2955},{"style":1228},[2956],{"type":49,"value":2957}," Risk ",{"type":44,"tag":458,"props":2959,"children":2960},{"style":1234},[2961],{"type":49,"value":2657},{"type":44,"tag":458,"props":2963,"children":2964},{"style":1228},[2965],{"type":49,"value":2966}," Likelihood ",{"type":44,"tag":458,"props":2968,"children":2969},{"style":1234},[2970],{"type":49,"value":2657},{"type":44,"tag":458,"props":2972,"children":2973},{"style":1228},[2974],{"type":49,"value":2975}," Impact ",{"type":44,"tag":458,"props":2977,"children":2978},{"style":1234},[2979],{"type":49,"value":2657},{"type":44,"tag":458,"props":2981,"children":2982},{"style":1228},[2983],{"type":49,"value":2984}," Mitigation ",{"type":44,"tag":458,"props":2986,"children":2987},{"style":1234},[2988],{"type":49,"value":2657},{"type":44,"tag":458,"props":2990,"children":2991},{"style":1228},[2992],{"type":49,"value":2993}," Status ",{"type":44,"tag":458,"props":2995,"children":2996},{"style":1234},[2997],{"type":49,"value":2657},{"type":44,"tag":458,"props":2999,"children":3000},{"style":1228},[3001],{"type":49,"value":3002}," Owner ",{"type":44,"tag":458,"props":3004,"children":3005},{"style":1234},[3006],{"type":49,"value":2685},{"type":44,"tag":458,"props":3008,"children":3010},{"class":1223,"line":3009},102,[3011],{"type":44,"tag":458,"props":3012,"children":3013},{"style":1234},[3014],{"type":49,"value":3015},"|---|---|---|---|---|---|---|\n",{"type":44,"tag":458,"props":3017,"children":3019},{"class":1223,"line":3018},103,[3020,3024,3029,3033,3038,3042,3047,3051,3055,3059,3064,3068,3073,3077,3081,3085,3089],{"type":44,"tag":458,"props":3021,"children":3022},{"style":1234},[3023],{"type":49,"value":2657},{"type":44,"tag":458,"props":3025,"children":3026},{"style":1228},[3027],{"type":49,"value":3028}," 1 ",{"type":44,"tag":458,"props":3030,"children":3031},{"style":1234},[3032],{"type":49,"value":2657},{"type":44,"tag":458,"props":3034,"children":3035},{"style":1228},[3036],{"type":49,"value":3037}," [specific risk tied to this design — not \"AI hallucination\" generically] ",{"type":44,"tag":458,"props":3039,"children":3040},{"style":1234},[3041],{"type":49,"value":2657},{"type":44,"tag":458,"props":3043,"children":3044},{"style":1228},[3045],{"type":49,"value":3046}," L\u002FM\u002FH ",{"type":44,"tag":458,"props":3048,"children":3049},{"style":1234},[3050],{"type":49,"value":2657},{"type":44,"tag":458,"props":3052,"children":3053},{"style":1228},[3054],{"type":49,"value":3046},{"type":44,"tag":458,"props":3056,"children":3057},{"style":1234},[3058],{"type":49,"value":2657},{"type":44,"tag":458,"props":3060,"children":3061},{"style":1228},[3062],{"type":49,"value":3063}," [specific control] ",{"type":44,"tag":458,"props":3065,"children":3066},{"style":1234},[3067],{"type":49,"value":2657},{"type":44,"tag":458,"props":3069,"children":3070},{"style":1228},[3071],{"type":49,"value":3072}," Done \u002F Planned \u002F Gap ",{"type":44,"tag":458,"props":3074,"children":3075},{"style":1234},[3076],{"type":49,"value":2657},{"type":44,"tag":458,"props":3078,"children":3079},{"style":1234},[3080],{"type":49,"value":1309},{"type":44,"tag":458,"props":3082,"children":3083},{"style":1240},[3084],{"type":49,"value":1314},{"type":44,"tag":458,"props":3086,"children":3087},{"style":1234},[3088],{"type":49,"value":1319},{"type":44,"tag":458,"props":3090,"children":3091},{"style":1234},[3092],{"type":49,"value":2864},{"type":44,"tag":458,"props":3094,"children":3096},{"class":1223,"line":3095},104,[3097],{"type":44,"tag":458,"props":3098,"children":3099},{"emptyLinePlaceholder":1258},[3100],{"type":49,"value":1261},{"type":44,"tag":458,"props":3102,"children":3104},{"class":1223,"line":3103},105,[3105,3109,3114,3118,3122,3127],{"type":44,"tag":458,"props":3106,"children":3107},{"style":1291},[3108],{"type":49,"value":1294},{"type":44,"tag":458,"props":3110,"children":3111},{"style":1297},[3112],{"type":49,"value":3113},"Residual risk after mitigations:",{"type":44,"tag":458,"props":3115,"children":3116},{"style":1291},[3117],{"type":49,"value":1294},{"type":44,"tag":458,"props":3119,"children":3120},{"style":1234},[3121],{"type":49,"value":1309},{"type":44,"tag":458,"props":3123,"children":3124},{"style":1240},[3125],{"type":49,"value":3126},"assessment",{"type":44,"tag":458,"props":3128,"children":3129},{"style":1234},[3130],{"type":49,"value":1251},{"type":44,"tag":458,"props":3132,"children":3134},{"class":1223,"line":3133},106,[3135],{"type":44,"tag":458,"props":3136,"children":3137},{"emptyLinePlaceholder":1258},[3138],{"type":49,"value":1261},{"type":44,"tag":458,"props":3140,"children":3142},{"class":1223,"line":3141},107,[3143],{"type":44,"tag":458,"props":3144,"children":3145},{"style":1234},[3146],{"type":49,"value":1491},{"type":44,"tag":458,"props":3148,"children":3150},{"class":1223,"line":3149},108,[3151],{"type":44,"tag":458,"props":3152,"children":3153},{"emptyLinePlaceholder":1258},[3154],{"type":49,"value":1261},{"type":44,"tag":458,"props":3156,"children":3158},{"class":1223,"line":3157},109,[3159,3163],{"type":44,"tag":458,"props":3160,"children":3161},{"style":1234},[3162],{"type":49,"value":1508},{"type":44,"tag":458,"props":3164,"children":3165},{"style":1273},[3166],{"type":49,"value":3167},"9. Recommendation\n",{"type":44,"tag":458,"props":3169,"children":3171},{"class":1223,"line":3170},110,[3172],{"type":44,"tag":458,"props":3173,"children":3174},{"emptyLinePlaceholder":1258},[3175],{"type":49,"value":1261},{"type":44,"tag":458,"props":3177,"children":3179},{"class":1223,"line":3178},111,[3180,3184,3189],{"type":44,"tag":458,"props":3181,"children":3182},{"style":1291},[3183],{"type":49,"value":1294},{"type":44,"tag":458,"props":3185,"children":3186},{"style":1297},[3187],{"type":49,"value":3188},"[APPROVED \u002F APPROVED WITH CONDITIONS \u002F CHANGES REQUIRED \u002F NOT APPROVED]",{"type":44,"tag":458,"props":3190,"children":3191},{"style":1291},[3192],{"type":49,"value":2640},{"type":44,"tag":458,"props":3194,"children":3196},{"class":1223,"line":3195},112,[3197],{"type":44,"tag":458,"props":3198,"children":3199},{"emptyLinePlaceholder":1258},[3200],{"type":49,"value":1261},{"type":44,"tag":458,"props":3202,"children":3204},{"class":1223,"line":3203},113,[3205,3209,3214],{"type":44,"tag":458,"props":3206,"children":3207},{"style":1291},[3208],{"type":49,"value":1294},{"type":44,"tag":458,"props":3210,"children":3211},{"style":1297},[3212],{"type":49,"value":3213},"Conditions (if any):",{"type":44,"tag":458,"props":3215,"children":3216},{"style":1291},[3217],{"type":49,"value":2640},{"type":44,"tag":458,"props":3219,"children":3221},{"class":1223,"line":3220},114,[3222,3227],{"type":44,"tag":458,"props":3223,"children":3224},{"style":1234},[3225],{"type":49,"value":3226},"-",{"type":44,"tag":458,"props":3228,"children":3229},{"style":1228},[3230],{"type":49,"value":3231}," [ ] [specific action before deployment — owner, deadline]\n",{"type":44,"tag":458,"props":3233,"children":3235},{"class":1223,"line":3234},115,[3236],{"type":44,"tag":458,"props":3237,"children":3238},{"emptyLinePlaceholder":1258},[3239],{"type":49,"value":1261},{"type":44,"tag":458,"props":3241,"children":3243},{"class":1223,"line":3242},116,[3244,3248,3253,3257,3262,3266,3271,3275],{"type":44,"tag":458,"props":3245,"children":3246},{"style":1291},[3247],{"type":49,"value":1294},{"type":44,"tag":458,"props":3249,"children":3250},{"style":1297},[3251],{"type":49,"value":3252},"Privacy review required?",{"type":44,"tag":458,"props":3254,"children":3255},{"style":1291},[3256],{"type":49,"value":1294},{"type":44,"tag":458,"props":3258,"children":3259},{"style":1228},[3260],{"type":49,"value":3261}," [Yes — run ",{"type":44,"tag":458,"props":3263,"children":3264},{"style":1234},[3265],{"type":49,"value":1237},{"type":44,"tag":458,"props":3267,"children":3268},{"style":1240},[3269],{"type":49,"value":3270},"\u002Fprivacy-legal:pia-generation",{"type":44,"tag":458,"props":3272,"children":3273},{"style":1234},[3274],{"type":49,"value":1237},{"type":44,"tag":458,"props":3276,"children":3277},{"style":1228},[3278],{"type":49,"value":3279},", if the plugin is installed \u002F\n",{"type":44,"tag":458,"props":3281,"children":3283},{"class":1223,"line":3282},117,[3284],{"type":44,"tag":458,"props":3285,"children":3286},{"style":1228},[3287],{"type":49,"value":3288},"No]\n",{"type":44,"tag":458,"props":3290,"children":3292},{"class":1223,"line":3291},118,[3293],{"type":44,"tag":458,"props":3294,"children":3295},{"emptyLinePlaceholder":1258},[3296],{"type":49,"value":1261},{"type":44,"tag":458,"props":3298,"children":3300},{"class":1223,"line":3299},119,[3301,3305,3310,3314],{"type":44,"tag":458,"props":3302,"children":3303},{"style":1291},[3304],{"type":49,"value":1294},{"type":44,"tag":458,"props":3306,"children":3307},{"style":1297},[3308],{"type":49,"value":3309},"Sign-off:",{"type":44,"tag":458,"props":3311,"children":3312},{"style":1291},[3313],{"type":49,"value":1294},{"type":44,"tag":458,"props":3315,"children":3316},{"style":1228},[3317],{"type":49,"value":3318}," [name, date]\n",{"type":44,"tag":458,"props":3320,"children":3322},{"class":1223,"line":3321},120,[3323],{"type":44,"tag":458,"props":3324,"children":3325},{"emptyLinePlaceholder":1258},[3326],{"type":49,"value":1261},{"type":44,"tag":458,"props":3328,"children":3330},{"class":1223,"line":3329},121,[3331],{"type":44,"tag":458,"props":3332,"children":3333},{"style":1234},[3334],{"type":49,"value":1491},{"type":44,"tag":458,"props":3336,"children":3338},{"class":1223,"line":3337},122,[3339],{"type":44,"tag":458,"props":3340,"children":3341},{"emptyLinePlaceholder":1258},[3342],{"type":49,"value":1261},{"type":44,"tag":458,"props":3344,"children":3346},{"class":1223,"line":3345},123,[3347,3351],{"type":44,"tag":458,"props":3348,"children":3349},{"style":1234},[3350],{"type":49,"value":1508},{"type":44,"tag":458,"props":3352,"children":3353},{"style":1273},[3354],{"type":49,"value":3355},"Cite check\n",{"type":44,"tag":458,"props":3357,"children":3359},{"class":1223,"line":3358},124,[3360],{"type":44,"tag":458,"props":3361,"children":3362},{"emptyLinePlaceholder":1258},[3363],{"type":49,"value":1261},{"type":44,"tag":458,"props":3365,"children":3367},{"class":1223,"line":3366},125,[3368,3373,3377,3381,3385,3389,3393,3397,3401,3406,3410,3415,3419],{"type":44,"tag":458,"props":3369,"children":3370},{"style":1228},[3371],{"type":49,"value":3372},"Regulatory citations in Section 6 (and anywhere else) were generated by an AI model and have not been verified against primary sources. Before the assessment is certified or relied on, run a verification pass against a legal research tool (Westlaw, EUR-Lex, or your firm's platform) for each cited provision — confirm the pinpoint, currency, and any delegated or implementing acts. The AI regulatory landscape shifts quickly; verify before advising. Source tags on each citation (e.g., ",{"type":44,"tag":458,"props":3374,"children":3375},{"style":1234},[3376],{"type":49,"value":1237},{"type":44,"tag":458,"props":3378,"children":3379},{"style":1240},[3380],{"type":49,"value":550},{"type":44,"tag":458,"props":3382,"children":3383},{"style":1234},[3384],{"type":49,"value":1237},{"type":44,"tag":458,"props":3386,"children":3387},{"style":1228},[3388],{"type":49,"value":292},{"type":44,"tag":458,"props":3390,"children":3391},{"style":1234},[3392],{"type":49,"value":1237},{"type":44,"tag":458,"props":3394,"children":3395},{"style":1240},[3396],{"type":49,"value":484},{"type":44,"tag":458,"props":3398,"children":3399},{"style":1234},[3400],{"type":49,"value":1237},{"type":44,"tag":458,"props":3402,"children":3403},{"style":1228},[3404],{"type":49,"value":3405},") show where it came from; ",{"type":44,"tag":458,"props":3407,"children":3408},{"style":1234},[3409],{"type":49,"value":1237},{"type":44,"tag":458,"props":3411,"children":3412},{"style":1240},[3413],{"type":49,"value":3414},"verify",{"type":44,"tag":458,"props":3416,"children":3417},{"style":1234},[3418],{"type":49,"value":1237},{"type":44,"tag":458,"props":3420,"children":3421},{"style":1228},[3422],{"type":49,"value":3423}," tags carry higher fabrication risk and should be checked first.\n",{"type":44,"tag":131,"props":3425,"children":3426},{},[3427,3432,3434,3439,3440,3445],{"type":44,"tag":135,"props":3428,"children":3429},{},[3430],{"type":49,"value":3431},"Before certifying the AIA (the Sign-off step, marking Status: APPROVED):",{"type":49,"value":3433}," Read ",{"type":44,"tag":62,"props":3435,"children":3437},{"className":3436},[],[3438],{"type":49,"value":604},{"type":49,"value":317},{"type":44,"tag":62,"props":3441,"children":3443},{"className":3442},[],[3444],{"type":49,"value":67},{"type":49,"value":3446},". If the Role is Non-lawyer:",{"type":44,"tag":394,"props":3448,"children":3449},{},[3450,3455,3463],{"type":44,"tag":131,"props":3451,"children":3452},{},[3453],{"type":49,"value":3454},"Certifying this AIA has legal consequences — it becomes the record the company relies on if a regulator or affected party asks how this use case was assessed. Have you reviewed this with an attorney? If yes, proceed. If no, here's a brief to bring to them:",{"type":44,"tag":131,"props":3456,"children":3457},{},[3458],{"type":44,"tag":458,"props":3459,"children":3460},{},[3461],{"type":49,"value":3462},"Generate a 1-page summary: the system, the regulatory classification, the risks identified, the mitigations in place, residual risk, open questions, what to ask the attorney before certifying.",{"type":44,"tag":131,"props":3464,"children":3465},{},[3466],{"type":49,"value":3467},"If you need to find an attorney, solicitor, barrister, or other authorised legal professional: your professional regulator's referral service is the fastest starting point (state bar in the US, SRA\u002FBar Standards Board in England & Wales, Law Society in Scotland\u002FNI\u002FIreland\u002FCanada\u002FAustralia, or your jurisdiction's equivalent).",{"type":44,"tag":131,"props":3469,"children":3470},{},[3471],{"type":49,"value":3472},"Do not proceed past this gate without an explicit yes. DRAFT assessments for attorney review do not require the gate — certification does.",{"type":44,"tag":120,"props":3474,"children":3475},{},[],{"type":44,"tag":124,"props":3477,"children":3479},{"id":3478},"risk-quality-standards",[3480],{"type":49,"value":3481},"Risk quality standards",{"type":44,"tag":131,"props":3483,"children":3484},{},[3485,3487,3492],{"type":49,"value":3486},"Same standard as the PIA skill — risks must be ",{"type":44,"tag":135,"props":3488,"children":3489},{},[3490],{"type":49,"value":3491},"specific and tied to the design",{"type":49,"value":212},{"type":44,"tag":1114,"props":3494,"children":3495},{},[3496,3517],{"type":44,"tag":1118,"props":3497,"children":3498},{},[3499],{"type":44,"tag":1122,"props":3500,"children":3501},{},[3502,3507,3512],{"type":44,"tag":1126,"props":3503,"children":3504},{},[3505],{"type":49,"value":3506},"Bad risk",{"type":44,"tag":1126,"props":3508,"children":3509},{},[3510],{"type":49,"value":3511},"Why bad",{"type":44,"tag":1126,"props":3513,"children":3514},{},[3515],{"type":49,"value":3516},"Better",{"type":44,"tag":1142,"props":3518,"children":3519},{},[3520,3538,3556],{"type":44,"tag":1122,"props":3521,"children":3522},{},[3523,3528,3533],{"type":44,"tag":1149,"props":3524,"children":3525},{},[3526],{"type":49,"value":3527},"\"AI hallucination\"",{"type":44,"tag":1149,"props":3529,"children":3530},{},[3531],{"type":49,"value":3532},"Applies to every LLM; says nothing",{"type":44,"tag":1149,"props":3534,"children":3535},{},[3536],{"type":49,"value":3537},"\"Model may generate plausible but incorrect legal citations — support agents have no current verification step before sending to customers\"",{"type":44,"tag":1122,"props":3539,"children":3540},{},[3541,3546,3551],{"type":44,"tag":1149,"props":3542,"children":3543},{},[3544],{"type":49,"value":3545},"\"Bias\"",{"type":44,"tag":1149,"props":3547,"children":3548},{},[3549],{"type":49,"value":3550},"Too vague",{"type":44,"tag":1149,"props":3552,"children":3553},{},[3554],{"type":49,"value":3555},"\"Résumé scoring model trained on historical hires; if historical cohort was demographically homogeneous, underrepresented candidates may be systematically scored lower\"",{"type":44,"tag":1122,"props":3557,"children":3558},{},[3559,3564,3569],{"type":44,"tag":1149,"props":3560,"children":3561},{},[3562],{"type":49,"value":3563},"\"Vendor risk\"",{"type":44,"tag":1149,"props":3565,"children":3566},{},[3567],{"type":49,"value":3568},"Circular",{"type":44,"tag":1149,"props":3570,"children":3571},{},[3572],{"type":49,"value":3573},"\"OpenAI's terms permit training on API inputs by default; unless the opt-out is confirmed in the agreement, customer support messages may be used to train the model\"",{"type":44,"tag":131,"props":3575,"children":3576},{},[3577],{"type":49,"value":3578},"Aim for 2-5 real risks, not 12 padded ones.",{"type":44,"tag":120,"props":3580,"children":3581},{},[],{"type":44,"tag":124,"props":3583,"children":3585},{"id":3584},"ai-policy-diff",[3586],{"type":49,"value":3587},"AI policy diff",{"type":44,"tag":131,"props":3589,"children":3590},{},[3591,3593,3598],{"type":49,"value":3592},"Every assessment should cross-check against the AI policy commitments in ",{"type":44,"tag":62,"props":3594,"children":3596},{"className":3595},[],[3597],{"type":49,"value":67},{"type":49,"value":3599},".\nCommon drift:",{"type":44,"tag":258,"props":3601,"children":3602},{},[3603,3615,3620,3625],{"type":44,"tag":56,"props":3604,"children":3605},{},[3606,3608,3613],{"type":49,"value":3607},"Policy prohibits AI use in ",{"type":44,"tag":458,"props":3609,"children":3610},{},[3611],{"type":49,"value":3612},"category",{"type":49,"value":3614}," — this use case is that category. Stop.",{"type":44,"tag":56,"props":3616,"children":3617},{},[3618],{"type":49,"value":3619},"Policy requires human review — this deployment has no human step. Design needs to change.",{"type":44,"tag":56,"props":3621,"children":3622},{},[3623],{"type":49,"value":3624},"Policy requires disclosure to affected parties — disclosure mechanism hasn't been built.",{"type":44,"tag":56,"props":3626,"children":3627},{},[3628],{"type":49,"value":3629},"Approved vendor list exists — this vendor isn't on it. Procurement step required.",{"type":44,"tag":131,"props":3631,"children":3632},{},[3633],{"type":49,"value":3634},"Flag every mismatch. One of them has to change before deployment.",{"type":44,"tag":120,"props":3636,"children":3637},{},[],{"type":44,"tag":124,"props":3639,"children":3641},{"id":3640},"handoffs",[3642],{"type":49,"value":3643},"Handoffs",{"type":44,"tag":258,"props":3645,"children":3646},{},[3647,3664,3682,3706],{"type":44,"tag":56,"props":3648,"children":3649},{},[3650,3655,3657,3662],{"type":44,"tag":135,"props":3651,"children":3652},{},[3653],{"type":49,"value":3654},"To product \u002F engineering:",{"type":49,"value":3656}," Conditions list with owners and deadlines. Not\n\"add oversight\" — \"add a human review step before any automated email is sent,\nowner: ",{"type":44,"tag":458,"props":3658,"children":3659},{},[3660],{"type":49,"value":3661},"product lead",{"type":49,"value":3663},", before launch.\"",{"type":44,"tag":56,"props":3665,"children":3666},{},[3667,3672,3674,3680],{"type":44,"tag":135,"props":3668,"children":3669},{},[3670],{"type":49,"value":3671},"To privacy:",{"type":49,"value":3673}," If personal data is involved, flag: \"Run ",{"type":44,"tag":62,"props":3675,"children":3677},{"className":3676},[],[3678],{"type":49,"value":3679},"\u002Fprivacy-legal:pia-generation [system name]",{"type":49,"value":3681}," in parallel, if the plugin is installed — the AIA doesn't substitute for a PIA.\"",{"type":44,"tag":56,"props":3683,"children":3684},{},[3685,3690,3692,3696,3698,3704],{"type":44,"tag":135,"props":3686,"children":3687},{},[3688],{"type":49,"value":3689},"To vendor-ai-review:",{"type":49,"value":3691}," If a new vendor is involved, flag: \"If there's no AI addendum reviewed for ",{"type":44,"tag":458,"props":3693,"children":3694},{},[3695],{"type":49,"value":2028},{"type":49,"value":3697},", run ",{"type":44,"tag":62,"props":3699,"children":3701},{"className":3700},[],[3702],{"type":49,"value":3703},"\u002Fai-governance-legal:vendor-ai-review",{"type":49,"value":3705}," before production.\"",{"type":44,"tag":56,"props":3707,"children":3708},{},[3709,3714],{"type":44,"tag":135,"props":3710,"children":3711},{},[3712],{"type":49,"value":3713},"To reg-gap-analysis:",{"type":49,"value":3715}," If new regulatory obligations emerged (EU AI Act high-risk, new sector rule), that skill tracks the gap.",{"type":44,"tag":120,"props":3717,"children":3718},{},[],{"type":44,"tag":124,"props":3720,"children":3722},{"id":3721},"close-with-the-next-steps-decision-tree",[3723],{"type":49,"value":3724},"Close with the next-steps decision tree",{"type":44,"tag":131,"props":3726,"children":3727},{},[3728,3730,3736],{"type":49,"value":3729},"End with the next-steps decision tree per CLAUDE.md ",{"type":44,"tag":62,"props":3731,"children":3733},{"className":3732},[],[3734],{"type":49,"value":3735},"## Outputs",{"type":49,"value":3737},". Customize the options to what this skill just produced — the five default branches (draft the X, escalate, get more facts, watch and wait, something else) are a starting point, not a lock-in. The tree is the output; the lawyer picks.",{"type":44,"tag":124,"props":3739,"children":3741},{"id":3740},"what-this-skill-does-not-do",[3742],{"type":49,"value":3743},"What this skill does not do",{"type":44,"tag":258,"props":3745,"children":3746},{},[3747,3752,3757,3762],{"type":44,"tag":56,"props":3748,"children":3749},{},[3750],{"type":49,"value":3751},"It doesn't approve the deployment. A human signs the assessment.",{"type":44,"tag":56,"props":3753,"children":3754},{},[3755],{"type":49,"value":3756},"It doesn't constitute any regulatory conformity assessment — where a regime (e.g., EU AI Act) requires a formal conformity assessment, that is a separate exercise requiring external legal review and technical documentation beyond what's here.",{"type":44,"tag":56,"props":3758,"children":3759},{},[3760],{"type":49,"value":3761},"It doesn't design the mitigations. It describes what needs mitigating; engineering\ndesigns the fix.",{"type":44,"tag":56,"props":3763,"children":3764},{},[3765],{"type":49,"value":3766},"It doesn't substitute for a PIA when personal data is involved. Run both.",{"type":44,"tag":3768,"props":3769,"children":3770},"style",{},[3771],{"type":49,"value":3772},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":3774,"total":3291},[3775,3789,3806,3813,3825,3836,3847],{"slug":3776,"name":3776,"fn":3777,"description":3778,"org":3779,"tags":3780,"stars":26,"repoUrl":27,"updatedAt":3788},"ai-inventory","track AI systems for EU AI Act","EU AI Act per-system inventory — track each AI system's role (provider, deployer, importer, distributor, authorized representative, product manufacturer) and risk tier (prohibited, high-risk, limited, minimal, GPAI, GPAI+systemic). Role and tier are assessed per system, not per company. Use when the user says \"ai inventory\", \"add an ai system\", \"what systems do we have\", \"classify this ai system\", \"eu ai act register\", or \"ai system registry\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3781,3784,3787],{"name":3782,"slug":3783,"type":16},"Compliance","compliance",{"name":3785,"slug":3786,"type":16},"Governance","governance",{"name":21,"slug":22,"type":16},"2026-05-14T06:02:19.677579",{"slug":3790,"name":3790,"fn":3791,"description":3792,"org":3793,"tags":3794,"stars":26,"repoUrl":27,"updatedAt":3805},"ai-tool-handoff","manage handoff to bulk legal review tools","Detects when Luminance, Kira, or a similar bulk-review tool is in use, hands off the high-volume clause extraction to it, and QAs its output per the trust level in `~\u002F.claude\u002Fplugins\u002Fconfig\u002Fclaude-for-legal\u002Fcorporate-legal\u002FCLAUDE.md`. Use when user says \"send to Luminance\", \"bulk review\", \"AI extraction\", or when diligence-issue-extraction hits a high-volume category.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3795,3798,3801,3802],{"name":3796,"slug":3797,"type":16},"Automation","automation",{"name":3799,"slug":3800,"type":16},"Contracts","contracts",{"name":21,"slug":22,"type":16},{"name":3803,"slug":3804,"type":16},"QA","qa","2026-05-14T06:01:31.00555",{"slug":4,"name":4,"fn":5,"description":6,"org":3807,"tags":3808,"stars":26,"repoUrl":27,"updatedAt":28},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3809,3810,3811,3812],{"name":21,"slug":22,"type":16},{"name":18,"slug":19,"type":16},{"name":14,"slug":15,"type":16},{"name":24,"slug":25,"type":16},{"slug":3814,"name":3814,"fn":3815,"description":3816,"org":3817,"tags":3818,"stars":26,"repoUrl":27,"updatedAt":3824},"amendment-history","trace contract amendment history","Trace how a contract has changed across its base agreement and all amendments — either a summary of all changes over time, or a provision trace for a specific clause. Use when the user says \"what changed in this contract over time\", \"show me the amendment history\", \"where's the latest [clause]\", \"how has [provision] evolved\", or uploads multiple versions of an agreement.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3819,3820,3823],{"name":3799,"slug":3800,"type":16},{"name":3821,"slug":3822,"type":16},"Documents","documents",{"name":21,"slug":22,"type":16},"2026-05-13T06:03:34.070339",{"slug":3826,"name":3826,"fn":3827,"description":3828,"org":3829,"tags":3830,"stars":26,"repoUrl":27,"updatedAt":3835},"auto-updater","check for community skill updates","Check installed community skills for updates. Shows a diff and requires explicit approval before applying. Use when the user says \"check for updates\", \"update my skills\", \"anything new for my installed skills\", or when invoked from the registry-sync agent.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3831,3832],{"name":3796,"slug":3797,"type":16},{"name":3833,"slug":3834,"type":16},"Plugin Development","plugin-development","2026-05-13T06:02:55.642269",{"slug":3837,"name":3837,"fn":3838,"description":3839,"org":3840,"tags":3841,"stars":26,"repoUrl":27,"updatedAt":3846},"bar-prep-questions","provide bar exam practice questions","Bar prep questions — MBE or essay, targeted at your weak subjects and bar jurisdiction. Tracks misses and comes back to patterns. Use when the user says \"bar prep\", \"MBE questions\", \"practice essay\", or \"test me for the bar\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3842,3845],{"name":3843,"slug":3844,"type":16},"Education","education",{"name":21,"slug":22,"type":16},"2026-07-24T05:41:43.01243",{"slug":3848,"name":3848,"fn":3849,"description":3850,"org":3851,"tags":3852,"stars":26,"repoUrl":27,"updatedAt":3861},"board-minutes","draft board and committee meeting minutes","Drafts board or committee meeting minutes in your house format. Auto-detects upcoming board and committee meetings from your calendar, asks for the agenda and any slides or pre-read materials, and produces a complete draft in the format learned from your seed minutes. Also handles written consents in lieu of meetings. Trigger: \"board minutes\", \"draft minutes\", \"upcoming board meeting\", \"committee minutes\", \"written consent\", or calendar detection of an upcoming board or committee event.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3853,3856,3857,3858],{"name":3854,"slug":3855,"type":16},"Documentation","documentation",{"name":3785,"slug":3786,"type":16},{"name":21,"slug":22,"type":16},{"name":3859,"slug":3860,"type":16},"Meetings","meetings","2026-05-14T06:01:29.792942",{"items":3863,"total":4046},[3864,3885,3899,3911,3930,3941,3960,3980,3994,4009,4017,4030],{"slug":3865,"name":3865,"fn":3866,"description":3867,"org":3868,"tags":3869,"stars":3882,"repoUrl":3883,"updatedAt":3884},"algorithmic-art","create algorithmic art with p5.js","Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3870,3873,3876,3879],{"name":3871,"slug":3872,"type":16},"Creative","creative",{"name":3874,"slug":3875,"type":16},"Design","design",{"name":3877,"slug":3878,"type":16},"Generative Art","generative-art",{"name":3880,"slug":3881,"type":16},"JavaScript","javascript",161831,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fskills","2026-04-06T17:56:15.455818",{"slug":3886,"name":3886,"fn":3887,"description":3888,"org":3889,"tags":3890,"stars":3882,"repoUrl":3883,"updatedAt":3898},"brand-guidelines","apply Anthropic brand colors and typography","Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3891,3894,3895],{"name":3892,"slug":3893,"type":16},"Branding","branding",{"name":3874,"slug":3875,"type":16},{"name":3896,"slug":3897,"type":16},"Typography","typography","2026-04-06T17:56:05.042852",{"slug":3900,"name":3900,"fn":3901,"description":3902,"org":3903,"tags":3904,"stars":3882,"repoUrl":3883,"updatedAt":3910},"canvas-design","create posters and visual art as PNG or PDF","Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3905,3906,3907],{"name":3871,"slug":3872,"type":16},{"name":3874,"slug":3875,"type":16},{"name":3908,"slug":3909,"type":16},"PDF","pdf","2026-04-06T17:56:03.794732",{"slug":3912,"name":3912,"fn":3913,"description":3914,"org":3915,"tags":3916,"stars":3882,"repoUrl":3883,"updatedAt":3929},"claude-api","build apps with the Claude API","Reference for the Claude API \u002F Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration.\nTRIGGER — read BEFORE opening the target file; don't skip because it \"looks like a one-liner\" — whenever: the prompt names Claude\u002FAnthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing\u002Fmodel choice\u002Flimits\u002Fcaching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent\u002FMCP\u002Ftool-definition\u002Fmulti-agent\u002FRAG\u002FLLM-judge\u002Fcomputer-use; generate\u002Fsummarize\u002Fextract\u002Fclassify\u002Frewrite\u002Fconverse over NL; debugging refusals\u002Fcutoffs\u002Fstreaming\u002Ftool-calls\u002Ftokens).\nSKIP only when another provider is being worked on (overrides all triggers): OpenAI\u002FGPT\u002FGemini\u002FLlama\u002FMistral\u002FCohere\u002FOllama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3917,3920,3921,3924,3926],{"name":3918,"slug":3919,"type":16},"Agents","agents",{"name":9,"slug":8,"type":16},{"name":3922,"slug":3923,"type":16},"Anthropic SDK","anthropic-sdk",{"name":3925,"slug":3912,"type":16},"Claude API",{"name":3927,"slug":3928,"type":16},"LLM","llm","2026-07-28T05:36:08.213335",{"slug":3931,"name":3931,"fn":3932,"description":3933,"org":3934,"tags":3935,"stars":3882,"repoUrl":3883,"updatedAt":3940},"doc-coauthoring","co-author documentation and technical specs","Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3936,3937],{"name":3854,"slug":3855,"type":16},{"name":3938,"slug":3939,"type":16},"Technical Writing","technical-writing","2026-04-06T17:56:14.18897",{"slug":3942,"name":3942,"fn":3943,"description":3944,"org":3945,"tags":3946,"stars":3882,"repoUrl":3883,"updatedAt":3959},"docx","create and edit Word documents","Use this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx files) or Word templates (.dotx files). Triggers include: any mention of 'Word doc', 'word document', '.docx', '.dotx', or requests to produce professional documents with formatting like tables of contents, headings, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx or .dotx files, inserting or replacing images in documents, performing find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a 'report', 'memo', 'letter', 'template', or similar deliverable as a Word or .docx file, use this skill. Do NOT use for PDFs, spreadsheets, Google Docs, or general coding tasks unrelated to document generation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3947,3948,3950,3953,3956],{"name":3821,"slug":3822,"type":16},{"name":3949,"slug":3942,"type":16},"DOCX",{"name":3951,"slug":3952,"type":16},"Office","office",{"name":3954,"slug":3955,"type":16},"Templates","templates",{"name":3957,"slug":3958,"type":16},"Word","word","2026-07-18T05:16:23.136271",{"slug":3961,"name":3961,"fn":3962,"description":3963,"org":3964,"tags":3965,"stars":3882,"repoUrl":3883,"updatedAt":3979},"frontend-design","design production-grade frontend interfaces","Guidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3966,3967,3970,3973,3976],{"name":3874,"slug":3875,"type":16},{"name":3968,"slug":3969,"type":16},"Frontend","frontend",{"name":3971,"slug":3972,"type":16},"React","react",{"name":3974,"slug":3975,"type":16},"Tailwind CSS","tailwind-css",{"name":3977,"slug":3978,"type":16},"UI Components","ui-components","2026-04-06T17:56:16.723469",{"slug":3981,"name":3981,"fn":3982,"description":3983,"org":3984,"tags":3985,"stars":3882,"repoUrl":3883,"updatedAt":3993},"internal-comms","write internal company communications","A set of resources to help me write all kinds of internal communications, using the formats that my company likes to use. Claude should use this skill whenever asked to write some sort of internal communications (status reports, leadership updates, 3P updates, company newsletters, FAQs, incident reports, project updates, etc.).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[3986,3989,3990],{"name":3987,"slug":3988,"type":16},"Communications","communications",{"name":3954,"slug":3955,"type":16},{"name":3991,"slug":3992,"type":16},"Writing","writing","2026-04-06T17:56:20.695522",{"slug":3995,"name":3995,"fn":3996,"description":3997,"org":3998,"tags":3999,"stars":3882,"repoUrl":3883,"updatedAt":4008},"mcp-builder","build MCP servers","Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node\u002FTypeScript (MCP SDK).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[4000,4001,4004,4005],{"name":3918,"slug":3919,"type":16},{"name":4002,"slug":4003,"type":16},"API Development","api-development",{"name":3927,"slug":3928,"type":16},{"name":4006,"slug":4007,"type":16},"MCP","mcp","2026-04-06T17:56:10.357665",{"slug":3909,"name":3909,"fn":4010,"description":4011,"org":4012,"tags":4013,"stars":3882,"repoUrl":3883,"updatedAt":4016},"read edit and manipulate PDF files","Use this skill whenever the user wants to do anything with PDF files. This includes reading or extracting text\u002Ftables from PDFs, combining or merging multiple PDFs into one, splitting PDFs apart, rotating pages, adding watermarks, creating new PDFs, filling PDF forms, encrypting\u002Fdecrypting PDFs, extracting images, and OCR on scanned PDFs to make them searchable. If the user mentions a .pdf file or asks to produce one, use this skill.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[4014,4015],{"name":3821,"slug":3822,"type":16},{"name":3908,"slug":3909,"type":16},"2026-04-06T17:56:02.483316",{"slug":4018,"name":4018,"fn":4019,"description":4020,"org":4021,"tags":4022,"stars":3882,"repoUrl":3883,"updatedAt":4029},"pptx","create and edit PowerPoint presentations","Use this skill any time a .pptx or .potx file is involved in any way — as input, output, or both. This includes: creating slide decks, pitch decks, or presentations; reading, parsing, or extracting text from any .pptx or .potx file (even if the extracted content will be used elsewhere, like in an email or summary); editing, modifying, or updating existing presentations; combining or splitting slide files; working with templates (.potx), layouts, speaker notes, or comments. Trigger whenever the user mentions \"deck,\" \"slides,\" \"presentation,\" or references a .pptx or .potx filename, regardless of what they plan to do with the content afterward. If a .pptx or .potx file needs to be opened, created, or touched, use this skill.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[4023,4026],{"name":4024,"slug":4025,"type":16},"PowerPoint","powerpoint",{"name":4027,"slug":4028,"type":16},"Presentations","presentations","2026-07-18T05:16:24.1471",{"slug":4031,"name":4031,"fn":4032,"description":4033,"org":4034,"tags":4035,"stars":3882,"repoUrl":3883,"updatedAt":4045},"skill-creator","create and optimize agent skills","Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[4036,4037,4038,4041,4044],{"name":3918,"slug":3919,"type":16},{"name":3854,"slug":3855,"type":16},{"name":4039,"slug":4040,"type":16},"Evals","evals",{"name":4042,"slug":4043,"type":16},"Performance","performance",{"name":3938,"slug":3939,"type":16},"2026-04-19T06:45:40.804",490]