California's new AI laws keep a human in charge of firing decisions: a standard leaders can adopt now

On 30 September 2026 the Office of California Governor Gavin Newsom announced that he had signed 13 bills on artificial intelligence. In employment, SB 947 (Senator Jerry McNerney) prohibits employers from relying solely on an automated decision system to discipline or dismiss a worker, and SB 951 (Senator Eloise Gómez Reyes) requires employers to tell workers when mass layoffs, relocations or terminations result from AI systems. AB 1331 and AB 1883 restrict workplace surveillance tools. In education, AB 2392 (Assemblymember Mike Fong) requires California's public higher-education institutions to set procurement standards for generative AI and to provide AI training for students, faculty and staff. Other measures keep doctors' judgement in charge when they use AI clinical tools, prohibit lawyers from fully delegating core legal work to AI, protect watermarks on AI-generated content, and update protections against deepfakes. "AI should expand opportunity – not come at the expense of workers and families," the Governor said. The Associated Press reported that the signing reverses the Governor's veto of a similar workplace bill in 2025, and Bloomberg Law reported that SB 947 takes effect on 1 July 2027.

This matters well beyond California. The state is home to most of the companies building frontier AI, and the size of its economy means its rules shape how national and international employers operate. One idea runs through the package: when a decision changes someone's life, whether that is their job, their diagnosis or their legal case, a person remains accountable for it. And it pairs that principle with a duty to teach, requiring AI training across the state's public universities rather than leaving literacy to chance.

Dan Fitzpatrick, The AI Educator, argues that California has written into law a principle good leaders already practise: outsource the doing, not the thinking. Automated tools can gather the evidence, flag the pattern and draft the letter, but the decision about a person belongs to a person who can be named and questioned. He sees the laws as policy catching up with practice: workers and teachers have been using AI ahead of their institutions, and the live questions now are governance, transparency and who gets taught to use these tools well. The wrong reading of this story is as a brake on adoption. The bigger risk for most organisations remains paralysis, and California's package describes how to adopt AI with structure: a human owner for consequential decisions, honesty with staff about which tools are in use, and training for everyone, not only the enthusiasts.

For school and college leaders, the practical step is an audit of every place an automated tool touches a decision about a person, from attendance and behaviour flags to admissions, recruitment and marking, with a named person responsible for each, and plain disclosure to staff of which tools are in use. For educators, the training duty in AB 2392 is a model worth copying at every level: AI literacy for students and staff alike, with verification of AI output taught as a core skill. For business leaders, it means an inventory of AI used in hiring, performance management and redundancy decisions, a rule that no such decision rests on a system alone, and clear notice to workers when AI changes their roles. Leaders outside California should not wait to be required to do this; it is what responsible adoption looks like.

Dan Fitzpatrick is a former teacher and senior school leader who now works with schools, governments and employers on adopting AI, and writes and speaks internationally on what AI means for learning, work and leadership. His approach is summed up as curiosity over fear, evolution over revolution, and humanity at the centre.

Dan Fitzpatrick is available for interview and comment on this story.