EU AI Act transparency and deepfake-labelling duties take effect
The rule survived a broader Digital Omnibus deal that pushed the Act's high-risk system obligations back to December 2027, leaving transparency as the deadline that actually arrived.
- Government & policy
- Major
Article 50 of the EU AI Act, covering transparency obligations, became enforceable on 2 August 2026. Providers of chatbots and other interactive AI systems must now tell users they are dealing with AI rather than a human, unless that would already be obvious to a reasonably well-informed person. Providers of systems that generate synthetic audio, image, video or text must mark their output in a machine-readable, detectable format, and deployers who publish or distribute deepfakes or AI-generated text on matters of public interest must disclose that the content was artificially generated or manipulated, with exceptions for clearly labelled artistic, satirical or editorially controlled work. More than 180 organisations had separately signed a voluntary Code of Practice on marking AI-generated content ahead of the deadline.
The date mattered partly for what it was not. The AI Act’s most consequential and contested provisions — obligations on “high-risk” systems listed in Annex III, covering areas such as employment, credit and law enforcement — had been due around the same period but were pushed back to December 2027 under a Digital Omnibus agreement reached earlier in the year, after industry and some member states argued the original timeline was unworkable. Article 50’s transparency and deepfake-labelling duties were not part of that delay and took effect as originally scheduled, alongside expanded enforcement powers for the European Commission’s AI Office over general-purpose AI models, including systems built on GPAI models within the same corporate group.
Enforcement rests mainly with national market surveillance authorities, with penalties for non-compliance running up to €15 million or 3% of global annual turnover. The 2 August milestone left the Act’s structure — its risk tiers, general obligations and penalty regime — intact, but demonstrated that the practical rollout of Europe’s flagship AI law was proceeding in stages, with the most demanding rules deferred and the more easily met transparency duties allowed to proceed on schedule.