The EU is getting serious about transparency rules: Starting August 2, 2026, the EU AI Act's mandatory labeling requirements come into force. Anyone publishing AI-generated content must clearly mark it as such – using watermarks, audio disclaimers, or embedded AI symbols. On June 10, the EU Commission published a voluntary code of conduct that specifies the technical criteria for this labeling.
Key Facts
- Deadline: August 2, 2026 – All synthetic content imitating real people or appearing deceptively authentic must be labeled
- Labeling formats: Permanently displayed AI symbols, watermarks, or audio disclaimers – marking must be visible but not intrusive
- Exceptions: Text under 200 tokens requires no watermark; deterministic algorithms and backend scripts are excluded
- Transition period: New AI systems must comply immediately; existing system providers get four months additional time
Staggered Implementation with Grace Period
The EU is taking a pragmatic approach: AI systems launched after August 2026 must comply from day one. For existing systems, there's a four-month grace period – they must adapt by December 2026. The labeling itself must be permanently visible but should not be intrusive. This aims to strike a balance between transparency and user experience.
Exceptions show the EU is thinking proportionally: Very short AI-generated text under 200 tokens doesn't need marking. Pure backend processes or deterministic algorithms without synthetic agents are also excluded.
High-Risk AI: Delays in Digital Omnibus
Parallel to the labeling requirement, the EU Parliament is debating the Digital Omnibus package, which delays deadlines for certain high-risk AI systems. Affected are AI applications in personnel selection, human resources, and creditworthiness assessment – these must meet strict requirements only from December 2027, with regulated products potentially getting until August 2028.
However, here's an important caveat: The core transparency rules remain untouched. Article 50 of the AI Act is not delayed by the Omnibus – meaning users must be informed from August 2026 when interacting with an AI agent, regardless of high-risk classifications.
Germany Prepares: AI Market Surveillance Act
On the national level, things are also moving: On June 10, the Bundestag's Digital Committee approved the AI Market Surveillance Act. It establishes the foundation for German oversight of the new EU rules and ensures German authorities can monitor compliance with the AI Regulation.
What This Means for German Companies
For German companies creating or distributing AI-generated content, August 2026 is a critical milestone. Those unable to demonstrate compliance with labeling requirements by then risk fines. The transition period for existing systems is generous but not unlimited – four months to upgrade should be sufficient if you start now. Companies in media, marketing, and creative industries especially should audit their AI tools and implement labeling mechanisms. The technical standards are now clear – uncertainty is no longer an excuse for delay.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
