AI Regulation: EU AI Act, US States, UK and Korea — 2026-09-04

Top developments

California Session Ends with 30 AI Bills Awaiting Governor’s Signature

On September 4, 2026, the Transparency Coalition reported that California legislators wrapped their 2026 session by passing 30 AI-related bills. Governor Gavin Newsom now has until September 30 to sign or veto these measures. This significant legislative output underscores California's role as a primary testing ground for US state-level AI regulation, particularly concerning chatbot disclosures and high-risk system obligations. The outcome of these signings will likely influence other states considering similar frameworks in the coming months.

EU AI Office Launches First Major Enforcement Sweep of Global AI Firms

According to a report by Korea Data Economy Newspaper published on September 2, 2026, the European Commission’s AI Office sent formal inquiries on August 26, 2026, to more than 30 major AI companies across the US, Europe, and Asia. The inquiries request details on safety measures, security protocols, and copyright compliance. This marks the first significant enforcement action following the August 2, 2026, deadline for Article 50 transparency obligations for general-purpose AI models. For Korean and global tech firms, this signals that the EU is actively moving from rule-making to active supervision of high-profile AI providers.

German Minister Proposes International AI Oversight Body

In a development reported by ZDF heute on September 4, 2026, German Minister Karsten Wildberger called for the creation of an international AI oversight authority modeled after the International Atomic Energy Agency (IAEA). This proposal aims to establish stricter global controls on AI models. While currently a policy suggestion rather than enacted law, it reflects growing concern among European policymakers about the pace of AI development outstripping national regulatory capabilities. If adopted, such a body could harmonize enforcement standards across jurisdictions, potentially reducing the compliance burden of navigating disparate EU, US, and Asian regulations.

Korea’s FTC Clarifies Verification Obligations for AI Product Claims

The Korean Fair Trade Commission (FTC) recently clarified regulations regarding advertising claims for AI and new technology products. As reported by Vision21 News on September 1, 2026, the FTC has made it clear that companies must provide empirical evidence (verification data) when advertising the performance of AI products. The new guidance also specifies conditions for extending submission deadlines for verification materials while shortening the allowable extension period. This move aims to prevent misleading marketing in the rapidly growing AI sector and ensures that consumer protection laws keep pace with technological claims.

Local view

Korea: Korea Deep published an analysis on September 3, 2026, examining the "Korea AI Ethics Principles" released earlier this year. The article highlights how the principles have shifted from abstract ideals to specific operational duties for providers, users, and the government. It notes that six requirements present in the 2020 draft have been removed, replaced by clearer role-based responsibilities. This reflects a maturing regulatory approach in Korea that balances innovation with ethical accountability.

Germany: Presse Control provided a detailed breakdown of the Digital Omnibus impact on German companies. The article emphasizes that while the EU AI Act became generally applicable on August 2, 2026, the Digital Omnibus (Regulation (EU) 2026/1744) has effectively delayed the strictest high-risk provisions until December 2027 and August 2028. This gives German enterprises a critical window to prepare for the most burdensome compliance requirements without facing immediate penalties for high-risk system non-compliance.

Context & numbers

• US State Laws in Force: As of late August 2026, comprehensive AI laws are effective or live in Texas (TRAIGA, Jan 1, 2026), Illinois, and Utah. Colorado's replacement framework is set for January 2027.
• California Compliance Timeline: California’s CPPA ADMT regulations are in force, with significant-decision obligations beginning January 1, 2027. SB 53 Frontier AI Transparency requirements are also advancing.
• EU High-Risk Delay: The Digital Omnibus has postponed Chapter III, Sections 1–3 of the EU AI Act (high-risk systems) to December 2027 and August 2028, respectively. However, transparency obligations (Article 50) for generative AI became enforceable on August 2, 2026.

On the radar

• September 30, 2026: Deadline for California Governor Gavin Newsom to sign or veto the 30 AI bills passed during the session.
• Q4 2026: Expect further guidance from the European Commission on the practical application of the Code of Practice for Transparency of AI-generated Content, which has seen approximately 190 companies sign up since its inception.
• Federal Preemption Watch: While the Senate struck the state AI law moratorium from the Big Beautiful Bill, federal preemption proposals remain a live legislative discussion, creating uncertainty for multi-state operators.