05/08/2026 - New obligations requiring AI systems to identify themselves and AI-generated content to be labelled have entered into force

The European Commission has started applying new transparency obligations under the Artificial Intelligence (AI) Act, requiring certain AI-generated content to be labelled and users to be informed when they are interacting with artificial intelligence, rather than a human. The rules, which entered into force on 2 August, represent another step in the EU’s effort to regulate the rapidly expanding use of AI across digital and audiovisual sectors.

Under the new requirements, chatbots and other interactive AI systems covered by the legislation must disclose that users are communicating with AI. AI-generated or -manipulated content, including deepfake images, videos and audio, must carry clear labels and machine-readable markers allowing it to be identified more easily. The obligations also cover AI systems used for emotion recognition and biometric categorisation, as well as AI-generated text published on matters of public interest where no human editorial review has taken place.

According to the Commission, the measures aim to help people distinguish authentic from AI-generated content and address growing risks linked to misinformation, fraud, impersonation and consumer deception. To support compliance, it has published guidelines explaining how providers and deployers can meet the new requirements, including through adherence to a voluntary Code of Practice on transparency of AI-generated content. The Commission has also published a first list of more than 180 organisations that have signed the code (see the list here).

Responsibility for overseeing the new rules is divided between several bodies. The European AI Office will supervise providers of general-purpose AI (GPAI) models as well as certain AI systems integrated into very large online platforms and search engines covered by the Digital Services Act. National competent authorities will enforce the rules for other AI systems, while the European Data Protection Supervisor will oversee their application within EU institutions, bodies and agencies.

The AI Office will also monitor compliance among providers of general-purpose AI models, which are now subject to requirements including technical documentation, copyright policies and the publication of sufficiently detailed summaries of training data. Providers of the most advanced models considered to pose systemic risks face additional obligations aimed at addressing threats including cybersecurity risks, harmful manipulation and loss of human control.

The new enforcement phase also covers the AI Act’s prohibitions on particularly harmful practices, including systems designed to manipulate people, exploit vulnerabilities or carry out certain forms of social scoring that threaten fundamental rights.

To support implementation, the Commission has launched reporting mechanisms allowing individuals and organisations to submit complaints about possible infringements, while people working with AI providers can use a secure whistleblower channel to report suspected violations. A dedicated channel has also been created for downstream providers using general-purpose AI models.

The AI Act will continue to be implemented gradually over the coming years. Following changes introduced through the AI Omnibus package, rules governing high-risk AI systems will apply from 2 December 2027, while requirements for high-risk AI systems integrated into regulated products will enter into force on 2 August 2028. New prohibitions covering AI systems used to generate non-consensual sexually explicit material and child sexual abuse material will apply from 2 December 2026.

Companies breaching the transparency requirements may face fines of up to €15 million or 3% of their global annual turnover, while EU institutions, bodies and agencies may be fined up to €750,000, with proportionality provisions applying to small and medium-sized companies.