The context
The AI Act entered into force on 1 August 2024. Regulation (EU) 2026/1744, in force since 27 July 2026, adjusted the implementation of certain rules while preserving the risk-based architecture.
The postponement primarily concerns obligations for high-risk systems. The original prohibitions, AI literacy, rules for general-purpose models and transparency obligations follow their own timetables.
What changed
New deadlines for high-risk systems
The rules for Annex III systems will apply from 2 December 2027. For systems integrated into products covered by Annex I, the date is 2 August 2028. Specific extensions exist for certain uses by public authorities.
Proportionate simplification
Some simplified measures previously reserved for SMEs have been extended to small mid-cap enterprises, including proportionate technical documentation and quality management, as well as greater access to sandboxes.
AI literacy reframed
Article 4 has applied since February 2025. Its amended wording focuses on measures to support the development of AI literacy. Recording training, the roles covered, risks and measures adopted remains a prudent way to demonstrate implementation.
What already applies
The prohibited practices originally set out in Article 5 and the AI literacy obligations became applicable on 2 February 2025.
The obligations in Chapter V have applied since August 2025, with specific rules for GPAI providers and models with systemic risk.
The transparency rules apply in August 2026, including informing people who interact with certain AI systems and marking certain synthetic content, subject to the statutory exceptions and transitional provisions.
Where the GDPR overlaps
Automated decisions
Where a decision is based solely on automated processing and produces legal effects or similarly significant effects, Article 22 GDPR may apply. The assessment requires consideration of the exceptions, human intervention and mechanisms for challenging decisions where provided for.
Special categories of data
Health data, biometric data used for unique identification and other categories under Article 9 require a specific condition for lawful processing. The AI Act does not create a general authorisation to process sensitive data.
Transparency towards individuals
Informing someone that they are interacting with AI does not replace GDPR information obligations concerning purposes, legal basis, recipients, retention, rights and, where applicable, the logic and consequences of automated processing.
Priorities for organisations
Identify chatbots, systems that interact directly with people and tools that generate or manipulate synthetic content.
Map proprietary and SaaS systems, their purpose, data, provider, legal role, risk level, oversight and available evidence.
For HR, credit, healthcare, education and other sensitive use cases, begin risk management, data governance, documentation, logging and human oversight before the deadline.
Official sources
- Regulation (EU) 2026/1744 — EUR-Lex.
- Regulation (EU) 2024/1689 — AI Act.
- European Commission — AI Act framework and timetable.
- Regulation (EU) 2016/679 — GDPR.
General information only. This content does not constitute a legal opinion, audit, certification or recommendation applicable without an assessment of the specific circumstances.