Methodology
This tool implements the classification rules of Regulation (EU) 2024/1689 (the EU AI Act) against the consolidated text published 2024-07-12, including the Digital Omnibus political agreement of 2026-05-22 that delayed the high-risk Annex III obligations to 2027-12-02.
What the classifier evaluates
- Article 5 prohibited practices. Any selected flag short-circuits the verdict to "prohibited" with a maximum fine ceiling of €35M or 7% of worldwide annual turnover (Article 99). Application date: 2025-02-02.
- Article 6 high-risk. Triggered by either being a safety component of an Annex I product OR falling into one of the 8 Annex III categories. Both together produce a dual classification covering Article 6(1) and Article 6(2).
- Article 50 transparency. Triggered when the system interacts with natural persons or generates synthetic content (audio / image / video / text).
- Articles 53 / 55 GPAI track. Orthogonal to the tier. Providers placing the GPAI model on the market trigger Article 53 obligations; cumulative training compute ≥ 1e25 FLOPs adds Article 55 systemic-risk obligations. Downstream integrators are mapped to Article 53(1)(b).
What it does NOT certify
This is a structured estimate, not legal advice. It does not assess Annex I harmonised product legislation conformity routes, post-market monitoring plans (Article 72), substantial-modification triggers, or national derogations under Article 2(3).
Sources
- Regulation (EU) 2024/1689 — consolidated text (Official Journal 2024-07-12)
- Annex III — categories of high-risk AI systems
- Implementing Acts 2025-2026 (AI Office guidance)
- Digital Omnibus political agreement — 2026-05-22 (Annex III deadline moved to 2027-12-02)
- European Commission AI Office published guidance (Q&A series 2025–2026)