EU AI Act: Chatbots must disclose AI identity from August 2
What it really says
On August 2, 2026, the transparency obligations under Article 50 of the EU AI Act take effect. From that date, three core obligations apply: First, providers of AI systems that interact directly with people (chatbots, voice assistants, conversational AI agents) must ensure users are informed no later than the first interaction that they are communicating with an AI. Second, providers of generative AI must ensure their outputs are marked with machine-readable labels enabling detection as AI-generated. Third, AI-generated or manipulated content depicting real persons realistically (deepfakes) must be clearly labeled as artificially generated or manipulated. Violations carry fines of up to 15 million euros or 3 percent of global annual turnover. Enforcement falls to national market surveillance authorities across all 27 member states. Importantly, while other high-risk obligations of the AI Act were deferred to December 2027 and August 2028 respectively through the Digital Omnibus, the Article 50 transparency obligations were explicitly excluded from the postponement. On July 20, 2026, the Commission published supplementary guidelines on the transparency obligations.
Our assessment
In five days, every chatbot in the EU must disclose that it is powered by AI, and deepfakes must be labeled as such. This is a direct response to one of the most common concerns about AI: that people are being manipulated without knowing they are dealing with a machine. The obligations are clearly formulated and apply to every chatbot provider operating in the EU, regardless of where the company is based. Realistically, many large providers like OpenAI, Google, and Anthropic already meet the disclosure requirement, as their chatbots typically introduce themselves as AI. The greater challenge lies with smaller providers and with the technical implementation of machine-readable marking for AI-generated content. For deepfake labeling, uniform technical standards are still lacking. The fact that other AI Act obligations were deferred while transparency rules were not shows the EU is deliberately prioritizing this area.
Relevance for Germany
For Germany, August 2 is a concrete date with immediate consequences. Every German company operating a chatbot, from savings banks to online shops, must ensure users learn at first interaction that they are speaking with an AI. The Federal Network Agency (Bundesnetzagentur) as the responsible market surveillance authority will monitor compliance. For German consumers, the regulation means greater clarity: anyone speaking with a customer service bot will know it going forward. For companies like DeepL, Aleph Alpha, or the AI startups in Berlin and Munich, it means specific compliance obligations for labeling their outputs. The deepfake labeling requirement is particularly relevant ahead of the 2029 federal election and given the increasing prevalence of AI-generated content on social media. Germany already has the institutional infrastructure for enforcement through the BSI and BNetzA.
Fact check
The transparency obligations under Article 50 of the EU AI Act are directly anchored in the legislative text and confirmed by the European Commission's official FAQ page dated July 20, 2026. The effective date of August 2, 2026 follows from the AI Act's timeline (Regulation 2024/1689). The fine ceiling of up to 15 million euros or 3 percent of annual turnover comes from Article 99 of the AI Act. The deferral of high-risk obligations through the Digital Omnibus (signed July 8, 2026) while maintaining Article 50 obligations is consistently documented by Greenberg Traurig, ComplianceHub, and the EU Artificial Intelligence Act portal. The three core obligations (chatbot disclosure, machine-readable marking, deepfake labeling) are consistently presented across all sources.
Source
- • https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act
- • https://www.gtlaw.com/en/insights/2026/6/deepfakes-chatbots-ai-generated-text-european-commission-details-transparency-obligations-under-the-ai-act
- • https://artificialintelligenceact.eu/transparency-rules-article-50/
- • https://compliancehub.wiki/eu-ai-act-article-50-transparency-digital-omnibus-2026/