Overview
- Article 50 took effect on August 2, 2026, requiring providers and deployers to clearly disclose when a person is interacting with an AI system such as a chatbot and to mark specified AI‑generated or AI‑manipulated content.
- The transparency duties cover machine‑readable labels for certain deepfakes and synthetic images, audio, video or text used on matters of public interest and also require notices when systems process biometric or emotion data.
- Enforcement is split across national market surveillance authorities, the EU AI Office for general‑purpose models and designated platforms, and the European Data Protection Supervisor for EU institutions, with fines up to €15 million or 3% of global turnover for breaches.
- Companies say the immediate burden is operational work such as mapping where AI is used, adding visible notices to interfaces, updating content workflows, and documenting compliance rather than removing AI features.
- The law’s transparency rules may shape global practice because firms often apply EU standards worldwide, even as the Act’s tougher high‑risk rules were delayed until late 2027 or 2028, a move criticised for leaving some vulnerable uses unregulated for the time being.