EU AI Act enforcement kicks in: AI Office gains supervision and fine powers from August 2
On August 14, Xinhua and China Mil reported that since August 2, the EU AI Act has expanded its scope, with the AI Office's supervisory and enforcement powers over general-purpose AI model providers formally taking effect, including document access, model evaluation, remediation orders, and fines.
On August 14, Xinhua and China Mil reported that since August 2, 2026, the EU AI Act entered substantive enforcement, with the AI Office's supervisory and enforcement authority over general-purpose AI model providers formally taking effect. The compliance obligations were live since August 2, 2025, but until August 2, 2026, the European Commission did not have direct enforcement power over GPAI providers.
Under the new regime, the AI Office gained full powers including technical document access, model evaluation, remediation orders, and fines. Three standing channels launched alongside: a public complaint tool, a downstream industry grievance channel, and an independent whistleblower tool. The Act requires interactive AI systems such as chatbots to disclose that users are interacting with AI rather than humans, and AI-generated or modified images, video, and audio deepfakes must be labeled and machine-readable. The European Commission published the first list of signatories to the AI Content Transparency Code of Conduct, covering more than 180 organizations, including OpenAI, which described its two complementary provenance systems: Content Credentials and SynthID watermarking.
Notably, the Digital Omnibus amendment pushed the compliance deadline for some standalone high-risk AI systems to December 2, 2027, while the prohibition on nudifier and non-consensual intimate-image systems moved forward to December 2, 2026, creating a divergent enforcement timeline.