Enforcement of the EU AI Act Begins

<p>Europe's AI law has transitioned from concept to enforcement.</p><p><strong>Why it matters:</strong> Brussels will assess its ability to implement AI regulations that companies can adhere to and whether it can establish a global benchmark for AI standards.</p><hr /><p><strong>Driving the news:</strong> A significant enforcement deadline occurred on August 2.</p><ul><li>New transparency and disclosure requirements are now in effect for chatbots and AI-generated content.</li><li>As of now, the EU's AI Office has not pursued any companies for misconduct under the law, but it can request information from companies or access to their models.</li></ul><p><strong>What they're saying:</strong> "This is a messy piece of legislation," stated Patrick Van Eecke, a partner at Cooley in Brussels who advises on AI Act compliance, in an interview.</p><ul><li>He noted that August 2 marks when ordinary users might start to notice the effects of the act, such as being informed they are viewing AI-generated material.</li></ul><p><strong>The big picture:</strong> The AI Act has the potential to become a global standard without being a direct copy for companies. While compliance may be complicated, the outcomes will influence how companies operate globally.</p><ul><li>"We need a standard. This technological moment is happening really quickly. And right now, at least the EU AI Act gives us a standard," said Amy Worley, managing director and data protection officer at Berkeley Research Group.</li><li>She mentioned that transparency notices required by the AI Act have already begun to spread, and watermarking could serve as a litigation defense for companies.</li></ul><p><strong>Case in point:</strong> Anthropic announced that future Claude models will generate text with a watermark to comply with the EU AI Act.</p><ul><li>According to <a href="https://www.anthropic.com/news/claude-text-watermark" target="_blank">Anthropic</a>, the watermarked text will be indistinguishable to readers, and it plans to offer a "watermark detection API" soon.</li><li><a href="https://blog.google/company-news/outreach-and-initiatives/public-policy/eu-ai-act-transparency-code-of-practice/?utm_source=chatgpt.com" target="_blank">Google</a> and Meta announced in July their intention to adopt and develop transparency and watermarking tools. <a href="https://about.fb.com/news/2026/07/meta-is-signing-the-eu-ai-act-code-of-practice-on-transparency-of-ai-generated-content/?utm_source=chatgpt.com" target="_blank">Meta</a> stated it would collaborate with the EU AI Office to ensure their compliance approach is "practical, interoperable and genuinely useful." </li><li><a href="https://help.openai.com/en/articles/12141645-eu-ai-act-openai-resources-and-customer-guidance?utm_source=chatgpt.com" target="_blank">OpenAI</a> is providing training data summaries in compliance with the Act, along with embedded "provenance signals." <a href="https://www.microsoft.com/en-us/trust-center/compliance/eu-ai-act?utm_source=chatgpt.com" target="_blank">Microsoft</a> is implementing internal governance and risk management changes to comply with the Act.</li></ul><p><strong>What's next:</strong> The transparency requirements that took effect in August are just one aspect of the AI Act's implementation.</p><ul><li>Regulations for high-risk AI systems, including applications in education, biometrics, and migration, as well as AI uses in physical products, will be introduced in December 2027 and August 2028.</li><li>This indicates that it will take years to fully evaluate the law's impact.</li></ul><p><strong>The bottom line:</strong> Europe may not compel the rest of the world to adopt its AI law verbatim, but it may not need to.</p><ul><li>If companies begin to view the AI Act as the default benchmark for transparency and risk management, Brussels will have influenced global AI regulations even where its law does not directly apply.</li></ul>