State AI Laws – Where Are They Now? // Cooley // Global Law Firm

The law firm Cooley has published an alert reviewing where several major US state AI laws now stand. It describes a turning point: many laws passed only a few years ago have been changed or postponed as 2026 compliance dates near, while federal action may reshape state efforts. It cites the White House's National Policy Framework for Artificial Intelligence, which urges Congress to preempt certain state laws, and California Executive Order N-5-26 of March 30, 2026, directing agencies to draft AI safety recommendations for state contractors.

Colorado's SB 205, enacted in May 2024, regulates high-risk AI in consequential decisions. Enforcement was postponed to June 30, 2026, and a working group draft from March 2026 would replace it with an automated decision-making law starting January 1, 2027. The draft narrows covered decisions, exempts low-risk uses, drops formal risk programs, impact assessments and incident reporting, keeps attorney general enforcement with a 90-day cure period, and removes the separate duty to avoid algorithmic discrimination. Passage remains unclear.

In California, AB 2013 on training data summaries took effect January 1, 2026 amid uncertainty about format and detail. SB 942 requires detection tools and disclosures from large generative tool providers; AB 853 moved its start to August 2, 2026 and added duties for large platforms from January 1, 2027 and device makers from January 1, 2028. Utah's SB 149, effective May 1, 2024, applies consumer protection law to AI and was narrowed in 2025.

New York's RAISE Act, signed December 2025, was amended on March 27, 2026 to follow California's frontier law more closely, though with penalties up to $1 million and $3 million and a 72-hour incident window against California's 15 days. The alert advises phased compliance planning.