J. Cholajda: AI Act – Entry into force of regulations concerning the labeling

AI Act – Entry into force of regulations on the labeling of deepfakes. New obligations for entities using artificial intelligence systems in the course of their professional activities.

On August 2, 2026, further provisions of Regulation (EU) 2024/1689 on artificial intelligence—the AI Act—took effect.
Under these provisions, chatbots and other interactive artificial intelligence systems will be required to inform users that they are interacting with artificial intelligence rather than a human.

Furthermore, content classified as “deepfakes”—images, videos, or audio generated or modified using artificial intelligence and resembling realistic objects—will have to be labeled by entities using AI in their professional activities. However, private individuals engaging in personal (and non-professional) activities are permanently exempt from this requirement.

Currently, these labels are to be applied by the entities using AI. According to the AI Act and the European Commission’s guidelines, these labels must be easily visible to the viewer and accessible during the first interaction with a deepfake, so as not to mislead the viewer regarding the synthetic origin of the material.

Failure to comply with these obligations may result in an administrative fine of up to 15,000,000 EUR or up to 3% of the total annual global turnover from the previous fiscal year, whichever is higher.

EU member states are also enacting their own legislation aimed at harmonizing national procedures with those of the AI Act. In Poland, the Act on Artificial Intelligence Systems, signed by the president, will soon come into force. It regulates the proceedings regarding violations of the AI Act, sets out the rules for imposing administrative fines for violations, and establishes a market supervisory authority for artificial intelligence.

The remaining regulations of the AI Act, including those regarding the labeling of deepfakes by providers of artificial intelligence systems, will not take effect until December 2, 2026, due to Regulation (EU) 2026/1744, adopted at the end of June 2026, which postponed the entry into force of certain provisions of the AI Act. The regulation, also known as Omnibus VII, deferred this deadline only for AI systems deployed before August 2, 2026. Providers of AI systems deployed after that date will have to comply with this obligation immediately.

By developing guidelines such as the Code of Conduct on Transparency of AI-Generated Content, the European Commission aims to facilitate compliance with the new obligations for the entities covered by them. This Code is not legally binding, yet its application ensures full compliance with the obligations set forth in Article 50 of the AI Act.