EU Starts Enforcing AI Act Rules on Deepfakes and Transparency
The EU began enforcing key AI Act provisions including deepfake identification and AI content labeling on August 2, 2026.
Why it matters: This enforcement milestone signals stricter AI regulation in Europe, affecting AI companies, legal tech firms, and corporate counsel globally. It sets a precedent for managing AI risks and enhancing transparency in AI deployments.
- Enforcement by EU Commission and national regulators commenced August 2, 2026.
- Duties include identifying deepfakes, labeling AI-generated content, and managing AI risks.
- Transparency rules under Article 50 require disclosing AI interactions and clear labeling of AI content.
- AI Office expanded with 38 staff and new tools; fines can reach 3% of global turnover.
On August 2, 2026, the European Commission's AI Office and national regulators activated enforcement of key provisions of the EU Artificial Intelligence Act. This marks the start of official oversight duties focused on identifying deepfakes, labeling AI-generated content, and managing risks associated with powerful AI systems.
The AI Act's transparency obligations, formalized in Article 50, now require AI providers to clearly label AI-generated material and disclose when users interact with AI systems. This effort aims to reduce deception risks and increase public trust. To assist compliance, the Commission published a voluntary Code of Practice in June 2026 and introduced EU icons for marking AI content.
The AI Office has been strengthened by adding 38 new staff members and introducing whistleblower and compliance tools for confidential reporting. The Office wields enforcement powers including information requests, access to AI models for evaluation, requirements to implement risk mitigation, and the ability to impose fines up to 3% of global annual turnover for non-compliance.
Since entering into force on August 1, 2024, the AI Act has seen phased implementation with general prohibitions from February 2, 2025, and now transparency measures beginning August 2, 2026. This staged approach underscores the EU's commitment to regulating AI technologies to protect fundamental rights while fostering innovation.
The Commission emphasized that these transparency requirements "help people recognise when content has been generated or altered by AI, reducing the risk of deception and manipulation." Enforcement dialogues will serve as a primary tool to assess compliance and clarify provider obligations.
By the numbers:
- 38 — new staff members added to the EU AI Office
- 3% — maximum fine of global annual turnover for violations
- August 2, 2026 — enforcement of transparency and risk oversight provisions began