Asger Risom
Enforcement of the EU AI Act’s transparency rules began on August 2, 2026, forcing platforms to label deepfakes and identify chatbots under threat of global turnover fines up to 3%.
As of August 2, 2026, the European Union has officially entered a new era of digital oversight as the European Commission began enforcing mandatory transparency requirements under its landmark Artificial Intelligence Act. Governed under Article 50 of the law, the regulations target conversational chatbots, deepfakes, and synthetic media to protect internet users from fraud, impersonation, and automated deception.
Clear warnings for chatbots and deepfakes
According to official documentation published by the European Commission, tech companies must now explicitly inform consumers whenever they interact with an automated agent rather than a real human. Interactive tools such as customer support bots, AI avatars, and virtual assistants must disclose their artificial nature immediately upon contact.
At the same time, any AI-generated or manipulated audio, video, or photorealistic images depicting existing people, places, or events must carry visible, standardized EU icons alongside invisible, machine-readable digital watermarks. AI-generated text published to inform the public on matters of general interest must also be clearly identified if it has not undergone human editorial review.
Massive fines for non-compliant platforms
To help companies adapt, the European Commission released guidelines and a voluntary Code of Practice on Transparency, which has already been signed by more than 180 major tech organizations. Enforcement is led jointly by national market surveillance authorities, the European Data Protection Supervisor, and the newly created European AI Office.
Failure to comply carries steep financial consequences. Companies caught violating the new transparency mandates risk fines reaching up to €15 million or 3% of their total global annual turnover, whichever is higher. While legacy generative AI systems placed on the market before August 2 receive a brief grace period until December 2026 to integrate machine-readable watermarks, basic labeling and chatbot disclosures apply immediately across the European single market.
