EU Begins Enforcing AI Act Transparency Rules

On August 2, 2026, the EU begins enforcing key provisions of the AI Act, requiring companies to disclose when users interact with AI and to label certain AI-generated or manipulated content. According to the European Commission, chatbots, AI agents, and digital humans must clearly inform users that they are interacting with an automated system rather than a human. Deepfake images, videos, and audio must be labeled, and covered content must include machine-readable markers for automatic detection.

These rules aim to reduce fraud, impersonation, misinformation, and other forms of deception, as synthetic content becomes increasingly difficult to distinguish from real material. The European Commission states that over 180 organizations have signed a voluntary Code of Practice to help companies apply transparency rules for AI-generated content.

The provisions taking effect include several specific obligations: banning certain manipulative systems, social scoring practices that exploit vulnerabilities or threaten individual rights; interactive systems must disclose AI identity; deepfake content must have visible labels; AI-generated or modified content must include technical markers; and AI-generated text involving public interest that has not been human-reviewed must be disclosed.

General-purpose AI model providers must document required information and provide it to regulators and companies building products on their systems; develop policies compliant with EU copyright law; publish sufficiently detailed training data summaries; and developers of the most powerful general-purpose models must address systemic risks such as cyberattacks, harmful manipulation, loss of control, and chemical, biological, radiological, or nuclear threats.

Enforcement responsibilities are divided among the European Commission's AI Office, national regulators, and the European Data Protection Supervisor. The AI Office will oversee general-purpose AI model providers, as well as certain AI services integrated into the largest online platforms and search engines covered by the Digital Services Act. National regulators are responsible for other AI systems within their jurisdictions, while the European Data Protection Supervisor oversees systems used by EU institutions.

The European Commission states that effective enforcement partly depends on whether member states appropriately designate and fund their national regulators. The AI Office has also introduced a whistleblower tool, allowing individuals and companies to submit complaints involving systems under the Office's supervision, and employees of AI companies can use a confidential reporting channel. There is also a separate reporting channel for companies relying on general-purpose models.

The AI Office can now enforce rules covering general-purpose AI models, which can perform a wide range of tasks and serve as the foundation for chatbots, agents, and other applications. More advanced models deemed to pose systemic risks face additional obligations, including assessing risks, implementing safeguards, and cooperating with regulators.

Regulators will receive technical support from a scientific panel of 60 independent AI experts, which recently held its first meeting. The AI Office has also appointed Alessandro Abate, a professor of computer science at the University of Oxford, as Chief Scientific Advisor to support model testing, safety, evaluation, innovation, and adoption.

Other parts of the AI Act will take effect later. The European Commission states that the AI Omnibus delays requirements for high-risk AI systems to December 2, 2027, while rules for high-risk AI embedded in regulated products are delayed to August 2, 2028. These systems may include AI used in areas with higher safety or rights risks.

The EU plans to introduce additional bans on December 2, 2026, including prohibiting systems that generate non-consensual sexual content or child sexual abuse material.