EU transparency rules for chatbots and synthetic media take effect

Article 50 transparency obligations under the EU AI Act started to apply on Aug. 2. The rules require providers to inform people when they are directly interacting with certain AI systems and to add machine-readable marking to covered synthetic or manipulated content. The legal materials reviewed do not provide a clock time, so this account does not assume the duties began at midnight.

Deployers also have disclosure duties for covered deepfakes, emotion recognition, biometric categorization and certain AI-generated public-interest text produced without human review or editorial control. The exact obligation depends on the system, content and role of the organization using it. A general label for all AI-assisted work would not explain those distinctions.

The European Commission's quick-facts page lists penalties of up to 15 million euros or 3 percent of worldwide annual turnover, subject to the legal framework and proportionality rules. That ceiling should not be presented as an automatic fine for every breach. Enforcement, exceptions and the facts of a particular case still matter.

Systems placed on the market before Aug. 2 receive a grace period for the machine-readable marking requirement in Article 50(2) until December 2026. The AI Omnibus also changed other deadlines, including some obligations for high-risk systems. It is therefore inaccurate to say that the entire AI Act became fully applicable on one date.

For publishers, the boundary around public-interest text and editorial control deserves close attention. Human review can affect the applicable disclosure duty, but it should be real review rather than a label added after automatic publication. Organizations need to map whether they are providers or deployers, which content is covered and how visible disclosure and machine-readable marking will be implemented.