Overview
- Article 50, which took effect on 2 August 2026, obliges providers and deployers to clearly tell people when they are interacting with an AI system and to label specified AI‑generated or AI‑manipulated images, audio, video, or text with machine‑readable marks where required.
- Enforcement is split between national market surveillance authorities, the EU AI Office for systems built on general‑purpose models and designated platforms, and the European Data Protection Supervisor when EU institutions are involved.
- Companies that fail to comply face fines of up to €15 million or 3% of global annual turnover, while EU bodies face separate, lower caps, making the rules a costly operational compliance requirement rather than a ban on AI use.
- The law distinguishes creators (providers) from users or hosters (deployers) and includes carve‑outs such as legal‑authority uses, human review, and editorial control that can change whether a disclosure or label is needed.
- Observers expect the rules to ripple beyond Europe as firms adopt common technical provenance tools and labels globally, and regulators in places like Canada are consulting on similar transparency measures that could raise enterprise governance burdens.