EU AI Act
In force (staggered)Regulation (EU) 2024/1689 on Artificial Intelligence (AI Act)
This applies to you if …
- You develop, distribute or operate AI systems in the EU.
- You use chatbots, AI-generated text, images, audio or video in customer-facing contexts.
- You build on general-purpose AI models as the basis of your own applications.
Core obligations
- Classify the risk level of the AI system (prohibited / high / limited / minimal).
- Transparency: mark AI-generated or -manipulated content in a machine-readable way (Art. 50).
- Disclose to users when they interact directly with an AI system (e.g. a chatbot).
- Ensure a sufficient level of AI literacy among your own staff.
- For high-risk AI (Annex III/I): full provider/deployer duties (risk management, data governance, documentation, human oversight) — under the Digital Omnibus these apply only from 02.12.2027 (Annex III) resp. 02.08.2028 (Annex I); the Art. 50 transparency obligations apply INDEPENDENTLY of this, already from 02.08.2026.
First steps
- Create an inventory of all AI systems and features in use.
- Determine and document the risk class of each system.
- Review chatbot and content-generation flows for labelling/disclosure duties.
- Assign internal responsibilities and evidence-keeping.
Keep reading — free
Unlock the full content
Enter and confirm your email address: you subscribe to the Signal Forge newsletter by FORGE and get instant access to this and all other registration-gated content. You can unsubscribe at any time.
Already registered? The link from your confirmation email unlocks this device again.
Deadlines
- 2025-02-02Bans on unacceptable-risk AI + AI-literacy obligations apply
- 2025-08-02Obligations for General-Purpose AI (GPAI) models apply
- 2026-08-02Art. 50 transparency obligations (AI labelling, chatbot/deepfake disclosure) apply — NOT postponed by the Digital Omnibus, they apply unchanged (a stand-alone regime, not part of the Chapter III high-risk obligations)
- 2027-12-02Obligations for stand-alone high-risk AI systems under Annex III apply — postponed by the Digital Omnibus (originally 02.08.2026); finally adopted 06/2026, amended by Regulation (EU) 2026/1744 (Official Journal 24.07.2026, in force since 27.07.2026)
- 2028-08-02Obligations for high-risk AI embedded in regulated products under Annex I apply — postponed by the Digital Omnibus (originally 02.08.2027); finally adopted 06/2026, amended by Regulation (EU) 2026/1744
Penalty for non-compliance
Tiered framework: up to EUR 35 million or 7 % of worldwide annual turnover ONLY for prohibited practices (Art. 5); for breaches of the Art. 50 transparency obligations the lower framework of up to EUR 15 million or 3 % of worldwide annual turnover applies (Art. 99); whichever amount is higher in each case.