Annex III Classification Changes and Updates
Follow topic LLM context A cited markdown file you can paste into your AI assistant (ChatGPT, Claude, a RAG or project knowledge base) to ground it in this topic. Contains: the overview, key law text, case law, enforcement and guidance for this topic. Everything links back to its source on overview.legal — legal information, not advice.The content specifically addresses amendments to Annex III, which represents a distinct regulatory mechanism for updating high-risk AI system classifications. This topic would capture the procedural and substantive aspects of how Annex III is modified over time.
Overview
5 sources · Jul 23, 2026Legal Framework
Annex III of the AI Act enumerates standalone high-risk AI systems—those classified as high-risk independent of their role as safety components in regulated products. The primary mechanism for updating Annex III is Article 7 of the AI Act, which empowers the European Commission to adopt delegated acts modifying, adding to, or removing use-cases from Annex III. This power is bounded by substantive criteria: the Commission must assess whether the AI system in question poses a significant risk of harm to health, safety, or fundamental rights, and whether such risk is not already adequately mitigated by existing Union law or the system's own design features.
Articles 103 and 104 of the AI Act illustrate the broader architecture of classification updates by amending sectoral regulations—specifically Regulation (EU) No 167/2013 (agricultural machinery) and Regulation (EU) No 168/2013 (two- and three-wheel vehicles). Each amendment inserts a new subparagraph requiring that, when delegated acts are adopted under those sectoral frameworks concerning AI systems functioning as safety components, the requirements set out in Chapter III, Section 2 of the AI Act must be taken into account. This creates a bridge between product safety legislation and the AI Act's high-risk regime.
Key Developments
No enforcement decisions or case law yet exist interpreting Annex III amendment procedures, as the AI Act's application dates have not fully arrived. However, the procedural design establishes several practical thresholds. The Commission's delegated act power under Article 7 is subject to the conditions of Article 8, meaning amendments must be based on documented evidence of emerging risks. The delegated acts are also subject to the standard scrutiny mechanisms under Regulation (EU) 2018/1725, allowing the European Parliament and Council to object. The interplay between Annex III updates and sectoral regulation amendments (as seen in Articles 103 and 104) signals that classification changes will likely proceed in parallel across both tracks—standalone high-risk categories and product safety component categories.
Practical Guidance
Monitor delegated act consultations: Article 7 amendments to Annex III will be preceded by Commission consultations; organizations deploying AI systems near current Annex III boundaries should track these to anticipate reclassification of their systems from non-high-risk to high-risk.
Map dual-track exposure: If your AI system functions as a safety component in machinery or vehicles regulated under Regulations 167/2013 or 168/2013, prepare for concurrent obligations under both the sectoral framework and AI Act Chapter III, Section 2, as mandated by Articles 103 and 104.
Maintain risk documentation aligned with Article 7 criteria: Since Annex III amendments hinge on demonstrated significant risk to health, safety, or fundamental rights, maintain internal risk assessments using these same criteria to anticipate whether your system may be swept into Annex III through future delegated acts.
Build classification flexibility into compliance programs: Because Annex III is explicitly designed to evolve, avoid static compliance architectures; ensure your AI governance framework can accommodate reclassification without full system redesign.