Operational guidance, not legal advice. This page distills named public sources (regulator guidance and industry practice). It is not a legal determination, not a notification decision, and not a substitute for your counsel, insurer, or a retained DFIR firm. Verify applicability and current deadlines for your facts and jurisdiction.
What is an AI coding governance checklist?
Updated
An AI coding governance checklist covers approved tools, provenance, review gates, secret and IP handling, and framework mapping. Rows map to ISO/IEC 42001, NIST SSDF, and NIST AI RMF. EU AI Act duties appear only where applicable. This page is not legal advice.
AI coding governance checklist, last verified 10 September 2026 against ISO/IEC 42001:2023, NIST AI RMF 1.0 (NIST AI 100-1), NIST SSDF SP 800-218 Rev. 1 and SP 800-218A, OWASP GenAI and LLM guidance, and Regulation (EU) 2024/1689 Articles 4 and 26 where AI-assisted development touches deployer or literacy duties. It is not legal advice, not certification, and does not determine that any regulation applies to YOU.
This is a governance checklist, not YOUR compliance file
Audience: a CTO, CISO, engineering leader, or compliance owner standing up controls for AI-assisted coding — IDE copilots, coding agents, and autonomous pull-request bots. This page is not legal advice. Checking a row is not a determination that ISO/IEC 42001 certification is required, that the EU AI Act applies, or that your programme is audit-ready.
ISO/IEC 42001:2023 is a voluntary AI management-system standard. NIST SSDF (SP 800-218 Rev. 1, with SP 800-218A for AI-augmented development) and NIST AI RMF 1.0 are U.S. guidance, not Union law. Regulation (EU) 2024/1689 applies only if your facts meet its scope — this page cites Articles 4 and 26 where relevant, not as a blanket coding-tool mandate. Last verified 10 September 2026. Not legal advice.
- Legal requirement versus framework guidance versus industry best practice: Union regulation articles bind only if they apply. ISO 42001 and NIST publications are frameworks — useful evidence, not statutes. OWASP and secure-SDLC practice are industry best practice unless your contract or sector rule makes them mandatory.
- The AI coding policy template guide on this site is the editable starting-point page. The detect-security-issues-ai-coding-agents guide on this site is the pipeline page. The EU AI Act overview on this site is the pillar page. The ISO 42001 framework guide on this site is the standard education page. The AI governance solution page on this site describes product positioning — not a certification.
- This page does not classify YOUR coding assistant as a high-risk AI system under Article 6. It does not run an EU AI Act applicability test.
Legal requirement versus framework guidance versus best practice
Label each control before you adopt it. A framework row is not automatically a legal duty, and a best practice is not automatically optional if your customer contract says otherwise. Last verified 10 September 2026. Not legal advice.
| Kind of text | Examples on this page | What it is not |
|---|---|---|
| Legal requirement — only if it applies | EU AI Act Article 4 AI literacy (applied from 2 February 2025 under Article 113(a)); Article 26 deployer duties for high-risk systems — only if YOU are a deployer and the system is high-risk. | Not a blanket mandate to govern every GitHub Copilot session. Not ISO 42001 certification. |
| Framework guidance | ISO/IEC 42001:2023 clauses on context, leadership, risk, and operation; NIST AI RMF Govern and Map functions; NIST SSDF PO, PS, PW, RV practices in SP 800-218 and AI-augmented notes in SP 800-218A. | Not Union law. Not a substitute for counsel on the EU AI Act or sector rules. |
| Industry best practice | OWASP LLM and GenAI risk categories applied to coding workflows; branch protection; secret scanning; human review before merge. | Not an audit opinion unless your auditor says so. Not a ShipReady Metrics certification. |
| ShipReady Metrics recommendation | Using the signed-in AI inventory, AI risk register, and AI-authored-code floor as tracking surfaces — described under Where this shows up in ShipReady Metrics. | Not an industry standard. Not legal advice. Not a classifier that proves code was AI-written. |
Master AI coding governance checklist
The table is the centerpiece. Each row is a control area for AI-assisted development. The ISO 42001 and NIST columns are crosswalk hints — not certification claims. Walk gaps with engineering and counsel. Last verified 10 September 2026. Not legal advice.
| Control area | Checklist item | ISO/IEC 42001 hint | NIST SSDF / AI RMF hint | EU AI Act — only if applicable | Kind of text |
|---|---|---|---|---|---|
| Governance and scope | Document which teams, repos, and environments may use AI coding tools; name an accountable owner. | Clause 5 leadership; Clause 4 context. | AI RMF Govern 1; SSDF PO.1 organisation preparation. | Article 4 literacy for staff who operate AI systems — Article 113(a), from 2 February 2025. | Framework guidance; EU row is legal requirement only if Act applies. |
| Approved tools | Maintain an allow-list of vendors, models, and IDE integrations; block unapproved endpoints. | Clause 8 operation — documented procedures. | SSDF PO.3 toolchain; AI RMF Map 1.1. | Not a specific Article — tool choice is organisational unless output feeds a regulated AI system. | Best practice; EU cell N/A unless counsel finds otherwise. |
| Data and IP | Rules on what code, secrets, customer data, and prompts may leave the boundary; retention and training-use flags. | Clause 6.1 actions to address risks; Annex A controls. | SSDF PW.1 design; OWASP LLM data-leakage category. | Article 26(2) deployer input-data relevance — only for high-risk deployers. | Best practice; EU row legal requirement only if high-risk deployer duty applies. |
| Provenance | Require commit metadata or PR labels that mark AI-assisted or agent-authored changes; keep prompts and agent logs per retention policy. | Clause 7.5 documented information. | SSDF PW.7 supply-chain records; SLSA provenance concept. | Not a specific coding-tool Article — provenance supports high-risk documentation if applicable. | Best practice. |
| Review gates | Risk-tiered human review before merge; no direct-to-main for agent PRs without checks. | Clause 8.1 operational planning. | SSDF RV.1 review; DORA change-management practice. | Article 14 human oversight — only for high-risk systems, not every pull request. | Best practice; EU row legal requirement only if system is high-risk. |
| Security testing | SAST, secret scan, dependency scan, and DAST on agent-touched paths before release. | Clause 8.1; risk treatment. | SSDF PW.4–PW.8; SP 800-218A AI-augmented code review. | Not a coding-tool-specific Article. | Industry best practice / SSDF guidance. |
| Exceptions | Time-boxed exception process with approver, scope, compensating controls, and expiry. | Clause 6.1 risk treatment; Clause 9 performance evaluation. | AI RMF Manage 2; SSDF RV.2. | Not applicable unless exception touches a regulated AI system. | Best practice. |
| Training | Role-based training on safe prompting, secret hygiene, licence risks, and when to escalate. | Clause 7.2 competence. | AI RMF Govern 2; EU AI Act Article 4 literacy concept. | Article 4 — providers and deployers shall take measures for AI literacy — from 2 February 2025. | Framework + EU literacy row is legal requirement only if Act applies. |
| Monitoring | Track AI-authored commit share, review bypass, incident rate on AI-touched files, and policy violations. | Clause 9 evaluation; Clause 10 improvement. | AI RMF Measure 1–4; DORA deployment and change-failure signals. | Not a coding-tool-specific Article. | Best practice; SRM can surface proxies — see product section. |
| Policy and cadence | Publish an internal AI coding policy; review at least annually or when tools change. | Clause 7.5; management review Clause 9.3. | SSDF PO.1; AI RMF Govern 1. | Article 26 deployer governance — only if high-risk deployer. | Best practice; policy template on this site is a starting point requiring legal review. |
EU AI Act — only where applicable
Most internal coding assistants are not, by themselves, high-risk AI systems under Article 6. Article 4 literacy and Article 26 deployer duties matter when the Act applies and YOU are a provider or deployer of an in-scope system. Last verified 10 September 2026 against Regulation (EU) 2024/1689 (OJ L 2024/1689, 12.7.2024). Not legal advice.
- Article 4: providers and deployers shall take measures to ensure a sufficient level of AI literacy — Chapter I, applied from 2 February 2025 under Article 113(a). That is a legal requirement only if the Act applies to YOU.
- Article 26: high-risk deployer duties include human oversight assignment and input-data relevance — only if YOU deploy a high-risk AI system. AI-generated application code in a non-high-risk product is not automatically Article 26.
- The EU AI Act overview on this site is the pillar page. The AI-governance-requirements guide on this site is the Articles 4, 14 and 26 page. This page does not determine applicability.
What you need to do now
Operational steps for engineering and compliance leaders. Not a filing, not certification, not legal advice. Last verified 10 September 2026.
- Inventory AI coding tools actually in use — IDE plugins, CLI agents, CI bots — and compare to your allow-list.
- Adopt or update an internal AI coding policy using the policy template guide on this site; have counsel review mandatory clauses for your jurisdiction and contracts.
- Turn on provenance: PR labels, commit trailers, or agent run IDs tied to change requests.
- Map your checklist gaps to ISO 42001 or NIST AI RMF only as evidence structure — not as a claim of certification.
- Ask counsel whether Article 4 literacy or Article 26 deployer duties apply to any AI system you ship, not merely to the coding tool.
- Wire security gates from the detect-security-issues-ai-coding-agents guide on this site before widening agent autonomy.
Checklist
Short verification list derived from the master table. Not YOUR audit file. Not legal advice.
- Named owner and written scope for AI-assisted coding.
- Allow-listed tools; unapproved tools blocked at network or MDM layer.
- Data/IP rules published; engineers trained on what must not enter prompts.
- Provenance on AI-assisted commits and agent PRs.
- Risk-tiered review — no silent merge to protected branches.
- SAST, secrets, and dependency scans on agent-touched code paths.
- Documented exception process with expiry.
- Annual policy review — or on material tool change.
- EU AI Act rows assessed with counsel — not assumed from this checklist.
Where this shows up in ShipReady Metrics
Honest mapping to shipped product surfaces only. None of the below is certification, legal classification, or proof that code was AI-generated with certainty.
- Signed-in Compliance → AI governance holds the AI inventory and AI risk register. You record which coding tools and models are in scope; the product does not classify systems under EU AI Act Article 6.
- The bundled eu_ai_act framework key on the obligation map is a starter control subset — illustrative, to be tailored. Marking it in-scope is not a determination that the Act applies.
- Signed-in AI Code Readiness reporting includes an AI-authored-code floor: a count from git markers or attributed-AI commits — not a statistical classifier.
- Signed-in Compliance → Policies hosts policy documents you upload or maintain. The product does not draft legal advice or certify ISO 42001.
- Security findings ingest (SAST, DAST, dependabot, secret scanning) supports the security-testing rows — described on the detect-security-issues guide; not linked here as an app route.
Primary sources (last verified 10 September 2026)
Every framework or regulatory claim on this page is taken from one of these. If a source revises, the date above shows we have not re-checked yet. Not legal advice.
ISO/IEC 42001:2023 — AI management system requirements (standard, not law). NIST AI RMF 1.0, NIST AI 100-1 — voluntary framework. NIST SP 800-218 Rev. 1 Secure Software Development Framework and NIST SP 800-218A — U.S. guidance including AI-augmented development notes. OWASP Top 10 and OWASP LLM/GenAI project materials — industry best practice. Regulation (EU) 2024/1689 Articles 4, 26 and 113 — legal requirements only if the Act applies; literacy from 2 February 2025 under Article 113(a). Commission AI Act pages are guidance, not the regulation.
Frequently asked questions
Is this checklist legal advice or ISO 42001 certification?
No. It is an operational checklist mapped to ISO/IEC 42001, NIST SSDF, NIST AI RMF, and EU AI Act articles only where applicable. Completing rows does not certify your organisation and does not determine that any regulation applies to YOU.
Does the EU AI Act require an AI coding governance programme for every Copilot user?
Not automatically. Article 4 literacy and Article 26 deployer duties apply only if the Act applies to YOUR role and systems. Internal coding assistance alone is not the same as deploying a high-risk AI system under Article 6. Ask counsel on YOUR facts.
Does ShipReady Metrics certify our AI coding governance?
No. The product tracks inventory, risk register entries, policy documents, and an AI-authored-code floor from git attribution — not ISO 42001 certification, not EU AI Act conformity, and not legal advice.
How does this relate to the policy template on this site?
The AI coding policy template guide is the editable starting point for the policy row in the master checklist. It requires legal review before adoption. This checklist is the control map; the template is the document shell.
Published by ShipReady Metrics, an evidence-based technology and compliance intelligence platform. This guide is educational and vendor-neutral.