Starting August 2: EU Chatbots Must Disclose AI Nature
What it really says
On August 2, 2026, the transparency obligations under Article 50 of the European AI Act take effect. This means three specific requirements: First, providers of AI systems that directly interact with people (such as chatbots) must clearly inform users that they are communicating with an AI, unless this is obvious from the circumstances. Second, providers of AI systems that generate synthetic content (images, audio, video) must mark this content as AI-generated in a machine-readable format. Third, AI-generated or AI-manipulated text published on matters of public interest must be labeled as such, with an exception where a human has editorially reviewed the text and bears editorial responsibility. Violations can result in fines of up to 15 million euros or 3 percent of global annual turnover, whichever is greater. Importantly, the EU Council approved the so-called Omnibus Regulation on June 29, 2026, which postpones the deadlines for high-risk AI systems from August 2026 to December 2027. However, Article 50 transparency obligations are not affected and take effect as planned on August 2, 2026.
Our assessment
Any company operating a chatbot in the EU must clearly label it as AI from August 2 onward. Violations carry fines of up to 15 million euros. This is a concrete response to the widespread concern that AI systems could deceive or manipulate people unnoticed. For the first time, a G7 jurisdiction enforces binding labeling requirements for chatbots and deepfakes with real penalties. For consumers, this represents a clear gain in transparency: if you are talking to an AI system, you should know it. At the same time, the regulation is pragmatic: where circumstances already make it obvious that an AI is involved (for example, a service clearly branded as an AI assistant), the obligation does not apply. Human-edited texts with editorial responsibility are also exempt. However, the postponement of high-risk deadlines through the Omnibus Regulation shows that the EU has yielded to industry pressure on more complex requirements. The transparency obligations still proceed on schedule, demonstrating that the EU is not compromising on fundamental consumer rights.
Relevance for Germany
For Germany, August 2 is a concrete deadline affecting many businesses directly. The Cologne Chamber of Commerce and Industry (IHK) is already actively warning about the new obligations: any company that uses a chatbot on its website, in its app, or in customer service must verify that its labeling meets the requirements. This affects not just tech corporations but also mid-sized companies with AI-powered customer service, online shops with chatbot advice, and media companies publishing AI-generated content. The German Federal Office for Information Security (BSI) has simultaneously proposed testing standard A5 for AI security, further strengthening Germany's regulatory framework. For small and medium enterprises, a simple, understandable label is sufficient in most cases according to the IHK. The national market surveillance authority in Germany will be responsible for enforcement. Particularly relevant: the obligation also applies to foreign providers whose AI systems reach EU users, providing protection for German consumers regardless of where the provider is headquartered.
Fact check
The transparency obligations under Article 50 of the EU AI Act are enshrined in the legislation and take effect on schedule on August 2, 2026. The Cologne IHK confirms this in its business advisory and specifies the concrete obligations for chatbot operators. The Omnibus Regulation, approved by the EU Council on June 29, 2026, postpones the deadlines for high-risk AI systems (Annex III) to December 2027 but explicitly leaves Article 50 transparency obligations unchanged. This is consistently confirmed by the Cloud Security Alliance, Holland & Knight, and Travers Smith. The fines of up to 15 million euros or 3 percent of global annual turnover are stipulated directly in the EU AI Act (Article 99). The exception for human-edited text with editorial responsibility is found in Article 50(4) of the AI Act.
Source
- • https://www.ihk.de/koeln/hauptnavigation/digitalisierung-und-innovation/digitalisierung/transparenzpflichten-nach-der-ki-verordnung-7100068
- • https://www.techtimes.com/articles/321174/20260721/eu-finalizes-ai-disclosure-rules-watermarking-mandate-outpaces-technology.htm
- • https://www.ad-hoc-news.de/wirtschaft/ki-transparenz-ab-august-neue-kennzeichnungspflichten-fuer-chatbots-und/69791178
- • https://cubbbix.com/blog/ai-regulation-july-2026-global-update/