EU to Begin Enforcing AI-Transparency Rules on Aug. 2, 2026

The European Commission said Friday that, starting Aug. 2, its AI Office and national authorities will begin enforcing the European Union’s AI Act transparency rules, a practical milestone that will be most visible to ordinary users. In plain terms, companies serving the EU market will have to make clear when people are interacting with AI and label AI-generated or AI-manipulated content so it can be identified as synthetic.

The requirements come from Article 50 of the EU AI Act, formally Regulation (EU) 2024/1689. The article says: “Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system.” It also says providers of AI systems, including general-purpose AI systems, that generate synthetic audio, image, video or text “shall ensure that the outputs of the AI system are marked in a machine-readable format and detectable as artificially generated or manipulated.” The rules are directly relevant to chatbots, generative AI tools, deepfakes and other synthetic media distributed in the EU.

The law allows for administrative fines for companies that do not comply. For breaches including the Article 50 transparency obligations, penalties can reach up to 15 million euros, or 3% of total worldwide annual turnover, whichever is higher. Enforcement is split between Brussels and national regulators: the Commission’s AI Office has the central role for providers of general-purpose AI models, while national competent authorities will lead most other cases in their own jurisdictions. The law’s highest penalty tier, for banned AI practices, is up to 35 million euros or 7% of worldwide annual turnover.

There is an important transition rule for some generative AI providers. Systems already on the market before Aug. 2, 2026, get a four-month grace period, until Dec. 2, 2026, to comply with the machine-readable marking requirement in Article 50(2). But systems placed on the market on or after Aug. 2 must comply immediately. The law also contains limited exceptions, including some law enforcement uses and certain artistic or satirical works.

This weekend is not the first day of the entire AI Act. The law entered into force on Aug. 1, 2024, and its prohibited-practices rules began applying on Feb. 2, 2025. Some governance and general-purpose AI provisions followed on Aug. 2, 2025. Still, Aug. 2, 2026, is the most visible compliance date so far for the public because it affects how companies present chatbots and AI-generated content. Henna Virkkunen, the European Commission’s executive vice-president for technological sovereignty, said in remarks quoted by The Associated Press: “As enforcement begins, we are taking an important step towards AI that people and businesses can understand and trust, and whose benefits are shared widely across our society.”

The Commission has spent recent weeks laying out how it expects companies to comply. It published guidelines on July 20 to help providers and deployers meet Article 50’s transparency duties, and on June 10 it released a voluntary code of practice on marking and labeling AI-generated content, including EU iconography for labels. Together, those documents are meant to give companies a clearer playbook as the bloc shifts from writing AI rules to enforcing them.