EU’s New AI Transparency Rules Take Effect in 2026 — Chatbots and Deepfakes Must Be Tagged
Published on 07/28/2026 at 20:42 | Redaktion boerse-global.de
Starting August 2, 2026, companies across the European Union will face binding transparency requirements for artificial intelligence systems, with the European Commission publishing detailed guidelines on how to label AI-generated content. The rules stem from Article 50 of the EU AI Act.
The legal framework gained teeth on July 27, when the “Digital Omnibus on AI” (Regulation (EU) 2026/1744) entered into force. While that regulation pushed back deadlines for high-risk AI systems, the general transparency obligations remain mandatory from August onward.
Operators of certain AI systems must make their use obvious to users. Chatbots, for instance, are required to disclose that people are interacting with a machine — deception is prohibited.
Deepfakes face equally strict labeling requirements. Any image, audio, or video content that realistically depicts people or events must be marked as AI-generated. The same rule applies to AI-written texts covering topics of public interest.
Labels must be both visible and machine-readable — digital watermarks or metadata are acceptable approaches. Existing systems have until December 2, 2026 to retrofit these markers.
Press and Art Exemptions
The regulation carves out protections for press freedom and artistic expression. AI-generated texts that undergo human editorial review, with a natural person taking responsibility, are exempt from labeling.
Works with an obvious artistic or satirical purpose benefit from lighter conditions. Standard image editing and purely fictional content without intent to deceive are also excluded.
Fines Reach Tens of Millions
In Germany, the Federal Network Agency (Bundesnetzagentur) will enforce the rules. At the EU level, the AI Office — staffed by 36 employees — oversees compliance. Penalties are steep: violations of transparency duties cost up to €15 million or 3 percent of global annual turnover.
Even tougher sanctions apply to banned practices, such as nudification apps, which will be prohibited from December 2026. Those violations carry fines of up to €35 million or 7 percent of worldwide revenue.
Companies Urged to Act Now
Experts recommend that businesses conduct an immediate inventory of their AI landscape. The critical question: does the company act as a provider or a deployer? Each role carries different obligations.
Providers are responsible for technically implementing watermarks. Deployers must ensure that notices are visible and understandable for end users. Specialists also advise promoting “AI literacy” among employees.
The timeline for remaining AI Act components is set. High-risk AI systems listed in Annex III must comply by December 2, 2027, while products covered under Annex I follow on August 2, 2028. At the national level, Germany’s state premiers agreed in late June on a Digital Media State Treaty (Digitaler-Medien-Staatsvertrag) to flesh out the EU requirements — expected to take effect around August 1, 2027.
Disclaimer regarding our articles: No investment advice, no buy or sell recommendation. Information on prices, companies, and markets is provided without guarantee; changes are possible at any time. Stock market transactions can lead to substantial losses. Our articles are created and reviewed in whole or in part automatically with the support of AI.
