AI Regulation: EU AI Act, US States, UK and Korea — 2026-10-03
California signed three AI workplace bills on September 30, Connecticut's sweeping AI law took effect October 1 with phased obligations through 2028, and the EU's AI Act transparency rules (Article 50) entered force August 2 with Commission enforcement now live. Meanwhile, federal preemption efforts in the US remain stalled, leaving state laws intact, while Korea prepares to clarify high-impact AI criteria by mid-2027.
AI Regulation: EU AI Act, US States, UK and Korea — 2026-10-03
Top developments
California Governor Signs Three AI Bills Targeting Workplace Surveillance (September 30, 2026)
On September 30, 2026, California Governor Gavin Newsom signed a package of artificial intelligence bills targeting the use of AI in the workplace. The bills address AI deployment in employment contexts and reflect California's continued momentum in AI regulation alongside its existing SB 53 (Frontier AI Transparency Act). This marks another layer of California's comprehensive state AI framework entering force.
Connecticut AI Law Phases In with October 1 Start Date
Connecticut's sweeping AI law, among the most wide-ranging state AI laws enacted, took effect October 1, 2026, with obligations phasing in through 2028. Companies that develop or deploy automated tools affecting Connecticut employees, consumers, or minors must begin assessing applicability immediately. The law addresses AI workforce development, social media use of minors, facial recognition technology, and more—making it one of the nation's most expansive state AI regimes.;
EU AI Act Article 50 Transparency Rules Now in Effect; Enforcement Begins
The European Union's AI Act transparency obligations under Article 50 entered force on August 2, 2026. These rules require providers and deployers of generative AI systems to disclose AI-generated content and inform users of AI interaction. Crucially, the Commission's enforcement actions—including requests for information, access to models, and model recalls—began on August 2, 2026, marking the start of active regulatory oversight. All providers of general-purpose AI systems, including those with synthetic audio, image, video, or text generation capabilities placed on the market before August 2, 2026, must have complied with Article 50(2) by that date.;
US Federal Preemption Bill Stalls; State Laws Remain in Force
As of September 2026, the bipartisan "Great American AI Act" federal preemption bill remains stalled since June, leaving state AI laws intact across the nation. Five states now have comprehensive AI laws in force or effective January 1, 2027: California, Colorado (which rewrote rather than repealed its AI Act), Texas, New York, and Illinois. No federal AI Act has been enacted, and the preemption fight continues to gridlock while state-level regimes proliferate.
Korea to Clarify High-Impact AI Standards by Mid-2027
South Korea's government is accelerating the development of guidelines for "high-impact AI" criteria, targeting January 2027 for initial completion and guaranteed publication by the first half of 2027. The new framework will clarify operator risk management and user protection duties in sectors such as healthcare and transportation, addressing regulatory uncertainty and field confusion. This represents Korea's AI Basic Act moving toward concrete enforcement rules.
Local view
Germany (Industrie-Fachwissen): The EU AI Act has applied since August 2, 2026, with immediate transparency and labeling obligations in force. High-risk AI duties, however, remain delayed until December 2, 2027, under the Digital Omnibus amendment. German companies are now required to develop AI competency, mark AI-generated outputs, and assess machine learning systems.
Germany (CloudMagazin): A commentary from Ivana Bartoletti (Wipro) emphasizes that governance structures, not arbitrary deadlines, matter most for EU AI Act compliance. With transparency rules now live and enforcement beginning, companies should focus on building durable governance rather than chasing deferred high-risk obligations.
France (Kohen Avocats): French legal guidance notes that since August 2, 2026, enterprises using AI must comply with Article 50 transparency obligations; penalties for non-compliance reach 7% of global turnover. AI-generated chatbots, resume screening systems, and intelligent video surveillance all fall under disclosure duties.
South Korea (AI Times): Researcher Choi Dae-sun (Soongsil University AI Safety Center) argues that government-backed "cybersecurity-specialized AI foundation models" fall into regulatory blind spots under Korea's AI Basic Act and require urgent criteria development.
Context & numbers
US state AI law spread: States passed 147 AI laws in 2025–26, with at least 30 reaching ordinary businesses. Five comprehensive state AI laws are now in force or effective January 1, 2027 (California, Colorado, Texas, New York, Illinois), alongside 40+ narrower laws on deepfakes, hiring, and chatbots.
Connecticut law scope: Connecticut's AI law applies to companies that develop or deploy automated tools affecting Connecticut employees, consumers, or minors—a broad trigger that catches most tech deployments.
EU enforcement timeline: All GPAI systems with synthesis capabilities released on or after August 2, 2025, must comply with transparency rules. The Commission began enforcement actions on August 2, 2026, giving providers a one-year grace period before active regulatory intervention.
On the radar
- Connecticut obligations phasing through 2028: Connecticut law obligations begin in phases through 2028; companies should conduct immediate impact assessments by Q4 2026 to map applicability.
- Korea high-impact AI criteria (H1 2027): South Korea's revised AI Basic Act implementation guidelines targeting January 2027 completion; watch for sector-specific standards in healthcare, transportation, and autonomous systems.
- EU AI Act high-risk delay (December 2, 2027): High-risk AI obligations remain deferred from the original August 2026 timeline to December 2, 2027, under the Digital Omnibus. Providers should begin high-risk assessments now to meet the extended deadline.
- Federal preemption stall continues: Absent movement on the Great American AI Act preemption bill, state-level fragmentation will persist through 2027; multi-state companies should plan compliance with at least five distinct state regimes.
This content was collected, curated, and summarized entirely by AI — including how and what to gather. It may contain inaccuracies. Crew does not guarantee the accuracy of any information presented here. Always verify facts on your own before acting on them. Crew assumes no legal liability for any consequences arising from reliance on this content.
Powered by
