One-line summary
The EU AI Act (Regulation (EU) 2024/1689), the world's first comprehensive law regulating artificial intelligence, entered into force on August 1, 2024 and becomes fully applicable with enforcement starting on August 2, 2026. The 12 Q&As below cover scope, bans, high-risk obligations, transparency rules, fines and export impact.
Basic Q&A
1. What is the EU AI Act?
The EU AI Act is the world's first comprehensive law regulating artificial intelligence. It regulates AI systems by risk tier, dividing applications into unacceptable risk (prohibited), high risk, limited risk and minimal risk, and entered into force on August 1, 2024.
2. When does the Act enter into force and become fully applicable?
The Act entered into force on August 1, 2024 and becomes fully applicable on August 2, 2026. Chapters I and II have applied since February 2, 2025; most rules including transparency apply from August 2, 2026; obligations for high-risk systems under EU sectoral safety legislation apply from August 2, 2027 (Article 113).
3. Who does the Act apply to?
It applies to providers, deployers, importers, distributors and authorised representatives of AI systems placed on or deployed in the EU market, regardless of where they are established. Any company whose AI output is used in the EU falls within scope, including Chinese exporters.
4. Which AI practices are prohibited?
Prohibited practices include social scoring, manipulative or exploitative techniques, and real-time remote biometric identification based on sensitive characteristics (strictly limited for law enforcement). Prohibition rules have applied since February 2, 2025; a new prohibition (Article 9, introduced by the AI Omnibus) takes effect in December 2026.
5. What is a high-risk AI system and when must it comply?
High-risk AI systems are those used in Annex III scenarios such as critical infrastructure, education, employment and law enforcement, and must meet risk management, data governance, logging and human oversight obligations. High-risk systems within EU sectoral safety legislation must comply from August 2, 2027; high-risk systems already in use before August 2027 and used at scale by public authorities must comply by January 2030 at the latest.
6. What do the transparency obligations require?
From August 2, 2026, chatbots and interactive AI systems must inform users that they are interacting with AI, and AI-generated or altered content (including deepfakes) must be clearly labelled. This obligation derives from Article 50, and the European Commission began enforcement on the same date.
7. What obligations apply to general-purpose AI (GPAI) models?
General-purpose AI models must meet transparency and copyright obligations; models with systemic risk face additional evaluation and risk-mitigation requirements. GPAI obligations have partially applied since August 2025 and are jointly enforced by the AI Office and national authorities from August 2, 2026.
8. What fines apply for non-compliance?
The maximum fine is EUR 35 million or 7% of a company's worldwide annual turnover, whichever is higher. Prohibited practices: up to EUR 35 million or 7%; GPAI obligations: up to EUR 15 million or 3%; other violations: up to EUR 7.5 million or 1% (Article 99).
9. How do the AI Act and ISO/IEC 42001 relate?
The AI Act is binding regulation while ISO/IEC 42001:2023 is a voluntary management system standard; they are complementary. Conformance with ISO/IEC 42001 can serve as evidence that an organisation meets some administrative requirements of the AI Act (e.g. risk management, records), but is not sufficient for full compliance.
10. What impact does the Act have on Chinese exporters?
Companies supplying AI products and services to the EU must comply by risk tier, including transparency labelling, high-risk system assessment and GPAI obligations. The Act has extraterritorial effect for output directed at the EU market; the Commission's AI Office is the main enforcement body.
11. What new requirements follow August 2, 2026?
Full enforcement starts, transparency rules (Article 50) apply, and GPAI and prohibition rules enter the enforcement phase. The European Commission's AI Office, together with national authorities, began enforcing the Act on that date; additional prohibitions take effect in December 2026.
12. Why is OpenAI watermarking ChatGPT text in the EU?
To comply with AI Act transparency obligations, OpenAI will watermark text generated by ChatGPT for EU users as required by the Act (reported by The Hindu on October 17, 2026). The watermark identifies AI-generated content and aligns with the Act's disclosure requirements for interactive AI systems and synthetic content.
Data Sources & References
- European Commission official AI Act page: Shaping Europe's digital future
- AI Act Service Desk, Article 113: Entry into force and application
- Xinhua: EU AI Act officially enters into force (2024-08-01): link
- European Commission press release: Commission starts enforcing AI Act rules and new transparency requirements on 2 August