Skip to content

Latest commit

 

History

History
82 lines (63 loc) · 10.3 KB

eu-ai-act.md

File metadata and controls

82 lines (63 loc) · 10.3 KB

EU AI Act: coverage matrix

Source: Regulation (EU) 2024/1689 (the Artificial Intelligence Act).

Policies live under international/eu_ai_act/v1/.

Legend: ✅ Implemented: checks real input fields against the obligation. ⚠️ Scaffold: package exists, returns placeholder denial. 📋 Planned: not in repo yet.

Title II: Prohibited AI practices (Article 5)

Obligation GOPAL policy Status Notes
Article 5(1)(a): Manipulative techniques prohibited_practices/manipulation ✅ Detects deployment of subliminal / manipulative techniques
Article 5(1)(a): Emotion recognition (workplace/edu) prohibited_practices/emotion_recognition ✅ Detects emotion-recognition systems in workplace/educational contexts
Article 5(1)(b): Vulnerability exploitation prohibited_practices/vulnerability_exploitation ✅ Article 5(1)(b) as a cumulative test: a vulnerability from age, disability or social/economic situation, plus behavioural distortion, plus significant harm
Article 5(1)(c): Social scoring prohibited_practices/social_scoring ✅ Detects social-scoring systems leading to detrimental treatment
Article 5(1)(d): Criminal-offense profiling prohibited_practices/criminal_profiling ✅ Prohibits prediction based solely on profiling; the Article 5(1)(d) carve-out requires both a supported human assessment and grounding in objective verifiable facts
Article 5(1)(e): Untargeted facial-recognition scraping prohibited_practices/facial_recognition_scraping ✅ Turns on untargeted collection from the internet or CCTV, not on facial recognition as such
Article 5(1)(b)+(g): Biometric categorization prohibited_practices/biometric_categorization ✅ Six sensitive attributes from Article 5(1)(g); exempts dataset labelling and law enforcement categorisation
Article 5(1)(h): Real-time remote biometric identification prohibited_practices/biometric_identification ✅ Closed list of Article 5(1)(h) objectives, and Article 5(3) prior authorisation required on top of a permitted objective

Title III: High-risk AI systems (Articles 6 – 27)

Chapter 2: Requirements for high-risk AI systems

Obligation GOPAL policy Status Notes
Article 9, Risk management system risk_management/risk_management ✅ Checks documented risk-management process & residual risk acceptance
Article 10, Data and data governance data_governance/data_quality ✅ Article 10(3) criteria are cumulative; Article 10(4) requires the deployment setting to be considered
Article 10, Training data data_governance/training_data ✅ Seven Article 10(2) practices; Article 10(5) special-category use conditional on safeguards
Article 11, Technical documentation documentation/technical_documentation ✅ Checks documentation.technical_documentation.completeness
Article 12, Record-keeping (logging) documentation/record_keeping ✅ Six-month floor from Articles 19 and 26(6); a longer sectoral period raises it, a shorter one cannot lower it
Article 12, Automated logs documentation/automated_logs ✅ Article 12(1)-(2) baseline, plus the four Article 12(3) fields for Annex III 1(a) biometric systems
Article 13, Transparency to deployers transparency/transparency ⚠️ Scores documentation completeness and a toxicity threshold. Article 13 requires instructions for use containing specified content; a completeness score is a proxy for that, not a test of it
Article 14, Human oversight human_oversight/human_oversight ✅ Article 14(4)(a)-(e) as four separate controls: oversight designed in, limits understood, automation bias, ability to disregard and to halt
Article 15, Accuracy, robustness, cybersecurity technical_robustness/robustness ✅ Accuracy declared in the instructions for use, feedback loops where the system keeps learning, and AI-specific cybersecurity attacks
Article 15, Fairness obligations eu_fairness/eu_fairness ✅ Checks bias-metric thresholds and protected-class coverage

Chapter 3: Obligations of providers, deployers, importers, distributors

Obligation GOPAL policy Status Notes
Article 16 and 17, provider obligations obligations/provider_obligations ✅ Ten Article 16 limbs tested separately, naming each one outstanding
Article 23, Importer obligations obligations/importer_obligations ✅ Article 23 pre-market verifications plus the ten-year retention duty
Article 24, Distributor obligations obligations/distributor_obligations ✅ Article 24 verifications plus the continuing Article 24(4) corrective duty
Article 26, Deployer obligations obligations/deployer_obligations ✅ Article 26, with 26(7) workplace and 26(11) affected-persons duties conditional, and 26(4) scoped to input data the deployer controls

Chapter 5: Conformity assessment and CE marking

Obligation GOPAL policy Status Notes
Article 43, Conformity assessment compliance/conformity_assessment ✅ Article 43 route selection: internal control for Annex III 1 only where harmonised standards were applied
Article 47, EU declaration of conformity compliance/declaration_conformity ✅ Article 47 requires machine readability, Annex V content, ten-year retention and translation
Article 48, CE marking compliance/ce_marking ✅ Article 48(2) digital marking and Article 48(4) notified body number, both conditional
Article 49, Registration in EU database compliance/registration ✅ Article 49(1) registration before market, and the Article 49(2) duty to register an Article 6(3) non-high-risk assessment

Title V: General-purpose AI models (Articles 51 – 56)

Obligation GOPAL policy Status Notes
Article 53, GPAI technical documentation gpai/technical_documentation ✅ Article 53(1)(a) with the Article 53(2) open-source exemption, which systemic risk removes
Article 53, Downstream transparency gpai/downstream_transparency ✅ Article 53(1)(b)-(d); the open-source exemption reaches the downstream duty only
Article 51, Systemic-risk classification gpai/systemic_risk_classification ✅ Article 51(2) 10^25 FLOP presumption, Article 52(1) notification, and the four Article 55 obligations

Not yet covered (📋 Planned)

Obligation Why it's open Help wanted
Article 6, High-risk classification rules Needs a structured input schema mapping Annex I/III to system attributes Yes
Article 27, Fundamental-rights impact assessment Needs FRIA template + checklist input Yes
Article 50, Transparency to natural persons (deepfakes etc.) Needs use-case taxonomy and disclosure-evidence input Yes
Article 55, Obligations for GPAI with systemic risk Needs systemic-risk evaluation results input Yes
Article 72, Post-market monitoring Needs post-deployment metrics input Yes
Annex IV, Technical documentation contents Currently aggregated into technical_documentation; could be split per-line Maybe

How to help

  1. Write a test for a case that has none. Every policy now has a sibling *_test.rego with an empty-input assertion, and scripts/check-test-coverage.sh fails CI if one goes missing. What is still thin is coverage of cases: a policy with one test proving it denies an empty input is verified against the fail-open class and nothing else. Adding a case needs no new Rego logic. See docs/tutorials/add-your-first-policy.md.
  2. Open a 📋 Planned row. Comment on the relevant issue (or open one) with your interpretation of the obligation and a proposed input schema before sending a PR.
  3. Disagree with our coverage call? Open an issue. We'd rather have the dispute in public than ship rules that misread the regulation.

⚠️ Reminder: GOPAL is not legal advice. The matrix above is GOPAL's engineering interpretation of where each obligation maps. Use it as a starting point for your own compliance review, not a substitute for one.