EU AI Act's Transparency Rules Are Now Live

The EU AI Act's Article 50 transparency obligations kicked in on August 2, 2026. Two things change in practice: any AI system interacting with a person has to disclose that it's an AI, and providers of generative systems must mark synthetic output in a machine-readable format — watermarks, metadata, or fingerprinting — so downstream tools can detect it. Deepfakes and AI-generated text published on matters of public interest need visible labels too.

The rules apply to anyone putting a system on the EU market, regardless of where they're headquartered. The Commission's Code of Practice on transparent generative AI is meant to give companies a presumption-of-conformity route, but it's still being finalized, which leaves providers implementing against a moving target. Penalties for non-compliance run up to €15M or 3% of global turnover.

The timing is awkward for the industry: this lands the same week ByteDance shipped Seedance 2.5, a video model generating 30-second clips with synced audio — exactly the class of output the marking rules exist for.

Why it matters: "is this AI?" stops being a UX choice and becomes a compliance surface with real fines attached.