EU Mandates Watermarks for AI Content: New Rules Take Effect August 2026
Published on 07/29/2026 at 13:12 | Redaktion boerse-global.de
Companies, media outlets and chatbot operators across the European Union face a hard deadline on 2 August 2026, when sweeping transparency rules for artificial intelligence come into force. The regulation, embedded in the Digital Omnibus on AI (Regulation EU 2026/1744), requires providers and professional users to clearly label computer-generated content — including deepfakes, text produced by language models and chatbot conversations.
The rules, which became law on 27 July 2026, impose two distinct layers of obligation. Providers of AI systems must ensure their output carries machine-readable markings, typically through digital watermarks or embedded metadata. Professional users — businesses, government agencies and news organisations — must then make those labels visible to the public whenever AI-generated text or deepfakes touch on matters of public interest.
Chatbot operators will also need to inform users explicitly that they are interacting with artificial intelligence, not a human.
Not all content falls under the mandate. Editorial text that has undergone human review is exempt, as is purely private use. Artistic and satirical works can skip the label if marking them would unreasonably interfere with the enjoyment of the work.
Fines of up to €15 million for non-compliance
Organisations that fail to comply risk penalties of up to €15 million or 3% of their global annual turnover, whichever is higher. To enforce the standards, the EU AI Office has received expanded powers, including the authority to conduct evaluations and access AI models directly. Those powers took effect on 27 July 2026 alongside the Digital Omnibus.
Existing AI systems get a grace period until 2 December 2026 to adapt to the labelling requirements. The same date marks the entry into force of a strict ban on nudification apps and systems that generate child sexual abuse material (CSAM).
A further regulatory milestone arrives on 2 December 2027, when the most stringent obligations for high-risk AI systems kick in. Those rules will introduce additional safety and control mechanisms for critical applications such as infrastructure management, law enforcement and employment decisions.
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.
