On 2 August 2026, the European Commission's AI Office, together with national authorities, began enforcing the Artificial Intelligence (AI) Act, including new transparency requirements that oblige certain AI systems to disclose when users are interacting with AI and when content has been generated or altered by it. The measures, announced in a press release dated 31 July 2026, require chatbots and other interactive AI systems to inform users they are dealing with AI, mandate labelling of deepfakes, and demand machine-readable marks on AI-generated content to facilitate detection. The Commission also published a first list of more than 180 organisations that have signed the Code of Practice on transparency of AI-generated content, operationalising the transparency rules. The enforcement aims to reduce deception and manipulation, help people make informed choices, and provide businesses with clearer obligations and a practical way to demonstrate compliance.
The AI Office can now enforce rules for providers of general-purpose AI (GPAI) models, including the most advanced models that may pose systemic risks, which face additional obligations to address risks of large-scale harm such as chemical, biological, radiological and nuclear incidents, loss of control, cyber offence, harmful manipulation, and threats to fundamental rights. All GPAI providers must document information for authorities and downstream providers, implement a copyright policy, and publish a detailed summary of training content. Enforcement also begins for prohibited AI practices, banning systems that manipulate people, exploit vulnerabilities, or unfairly score individuals. Responsibility for enforcement is shared: the AI Office handles systems offered by the same provider as the underlying GPAI model and those integrated into very large online platforms designated under the Digital Services Act; national authorities enforce rules for other systems; and the European Data Protection Supervisor covers EU institutions' AI use. Effective enforcement depends on member states designating and resourcing national authorities.
Scientific support comes from the Scientific Panel, an expert advisory body of 60 independent AI experts, which recently held its first meeting. The AI Office has appointed Professor Alessandro Abate of the University of Oxford's Department of Computer Science as Lead Scientific Adviser to support work on GPAI models, including innovation, adoption, testing, and evaluation. To facilitate enforcement, the AI Office launched a Complaint Tool for reporting alleged infringements, a Whistleblower Tool for secure reporting by individuals working with providers, and a dedicated channel for downstream providers to report infringements by GPAI model providers; information received is treated confidentially.
The AI Omnibus postponed application of rules on high-risk AI systems to 2 December 2027, and for high-risk systems integrated into regulated products to 2 August 2028. It also introduces new prohibitions on AI systems generating non-consensual sexually explicit content and child sexual abuse material, applying from 2 December 2026. Executive Vice-President Henna Virkkunen stated that the AI Act establishes a clear, risk-based framework for trustworthy AI, giving innovators legal certainty while protecting the public interest, and that enforcement marks an important step towards AI that people and businesses can understand and trust.
The enforcement imposes new compliance costs on AI providers, particularly GPAI developers who must document training data, implement copyright policies, and publish summaries, while transparency labelling affects chatbot operators and content platforms. Consumers benefit from reduced deception and manipulation, gaining clearer information to make informed choices. National authorities face administrative burdens in designating and resourcing competent bodies, while the AI Office's expanded role strengthens EU-level oversight, potentially shifting power from member states to Brussels. Businesses using AI in high-risk sectors face delayed obligations until 2027-2028, providing temporary relief but prolonging uncertainty. The measures aim to balance innovation with public safety, but the additional requirements may disproportionately affect smaller AI developers lacking resources for compliance.