On 2 August 2026, the EU AI Act's transparency rules became enforceable, requiring chatbots to disclose their AI nature and AI-generated media to carry machine-readable marks. OpenAI, Anthropic, and Google must ensure compliance for their models, ChatGPT, Claude, and Gemini, respectively, or risk fines up to €15 million or 3% of global annual turnover.
TL;DR
- EU AI Act's transparency rules are now enforceable, affecting chatbots like ChatGPT, Claude, and Gemini.
- Non-compliance with the EU AI Act can result in fines up to €15 million or 3% of global annual turnover.
- OpenAI, Anthropic, and Google have taken steps to comply, but specific details on implementation are scarce.
What happened
The EU AI Act's transparency rules under Article 50 became enforceable on 2 August 2026. These rules require chatbots to disclose their AI nature and AI-generated media to carry machine-readable marks.
OpenAI, Anthropic, and Google must ensure their models, ChatGPT, Claude, and Gemini, respectively, comply with these rules or face fines up to €15 million or 3% of global annual turnover.
OpenAI has begun applying invisible watermarking to images generated through ChatGPT, Codex, and the API. Google, the company behind SynthID watermarking technology, is in a strong technical position for the marking obligation. Anthropic has not provided specific details on its compliance measures.
Why it matters
For AI/ML developers and startup founders, compliance with the EU AI Act is crucial to avoid hefty fines and maintain trust with users.
Tech investors should consider the compliance status of AI models when evaluating startups and companies in the EU market.
The competitive landscape may shift as companies prioritize compliance and transparency, potentially affecting market share and user preferences.
Key facts
- EU AI Act transparency rules became enforceable on 2 August 2026.
- Fines for non-compliance can reach up to €15 million or 3% of global annual turnover.
- OpenAI has applied invisible watermarking to images generated through ChatGPT, Codex, and the API.
- Google's SynthID watermarking technology puts Gemini in a strong technical position for compliance.
- Anthropic has not provided specific details on Claude's compliance measures.
- The EU AI Act applies to any chatbot accessible by EU users, including those based in Ireland.
- Deployers of AI models must also ensure compliance, adding their own visible AI disclosures and updating terms of service.
Context
The EU AI Act follows a risk-based approach, classifying AI systems into four different risk categories. Article 50 covers 'limited-risk' AI systems, including chatbots and content generators.
The Act's transparency obligations require providers and deployers to inform users they are interacting with an AI system and to mark AI-generated media with machine-readable labels.
Enforcement of the Act is split between national authorities and Brussels, with fines up to €15 million or 3% of global annual turnover for non-compliance.
