California on June 25 became the first state in the country to deploy a systematic, publicly available tool for tracking whether artificial intelligence is displacing workers — and the initial findings offer both reassurance and early warning signs that complicate the narrative on both sides of the AI labor debate.
The California AI-Unemployment Tracker, developed in partnership between UCLA’s California Policy Lab and the state’s Employment Development Department, cross-references unemployment insurance claims against occupation-level AI exposure scores to detect patterns that might signal technology-driven job loss before it reaches crisis scale. The dashboard will be updated monthly, and the underlying methodology and data are publicly available — a transparency decision that allows researchers, employers, labor organizations, and policymakers outside Sacramento to run their own analyses.
What the Data Shows — and What It Doesn’t
The headline finding is that there is currently no evidence of rising statewide unemployment claims in occupations classified as having high AI exposure. That finding pushes back against the more alarmist projections that have circulated since generative AI tools entered the mainstream in late 2022.
But the data is not uniformly reassuring. Dr. Ben Hyman, a senior researcher at the California Policy Lab, noted that while large-scale AI-related layoffs are not visible in the statewide numbers, the tracker does reveal patterns in specific populations. Claims from college-educated workers in occupations with high AI exposure increased after the public release of ChatGPT-3.5 in November 2022. Workers in high-exposure roles concentrated in the San Francisco Bay Area experienced a sustained increase in unemployment claims that has persisted into 2026. And the data shows these effects are appearing in tech-heavy sectors where AI adoption has been fastest and most aggressive.
The nuance matters. A statewide average that shows no displacement can obscure regional and demographic concentrations where the impact is real and measurable. The tracker is designed to surface exactly those concentrations before they compound into structural workforce problems.
One finding that may ease concerns about inequitable impact: the initial analysis did not detect large disproportionate increases in AI-exposed unemployment claims by race, ethnicity, gender, or age. That does not mean demographic disparities will not emerge as AI adoption accelerates, but it does suggest the current wave of displacement — to the extent it exists — is hitting along occupational and geographic lines rather than demographic ones.
Why California Built It
The tracker is a direct product of Governor Newsom’s March 2026 executive order on AI and the workforce — his third executive order on artificial intelligence since 2023, and the first focused specifically on labor market preparation. The order directed state agencies to work with universities, labor economists, and industry leaders to develop early warning systems for workforce disruption and to ensure that workers share in the productivity gains AI generates.
Stewart Knox, Secretary of California’s Labor and Workforce Development Agency, framed the tool in operational terms. The tracker provides a clearer picture of how AI is affecting working people and jobs, he said, and allows the state to ground its decisions in data rather than speculation — identifying where to focus retraining, job-search support, health-coverage guidance, and other transition services before displaced workers fall through the gaps.
Till von Wachter, professor of economics at UCLA and faculty director of the California Policy Lab, acknowledged the anxiety directly. Workers’ concerns about what AI could mean for their jobs are real, he said. The tracker is designed to replace speculation with evidence.
EDD Director Nancy Farias described the tool as a mechanism for connecting Californians with resources, training, and support in real time — not after the fact. The distinction between reactive and proactive intervention is central to the tracker’s design. Traditional labor market data, including monthly jobs reports and quarterly unemployment surveys, operates on a lag that can leave displaced workers without support for months. The AI-Unemployment Tracker is intended to shorten that gap by flagging emerging patterns in claims data before they appear in headline employment statistics.
The Policy Infrastructure Behind the Dashboard
The tracker does not exist in isolation. California has assembled a layered regulatory and workforce development framework around AI over the past three years that no other state has replicated at comparable scale.
The Transparency in Frontier Technology Act, signed in September 2025 as Senate Bill 53, established the first state-level legislation requiring disclosure and safety protocols from developers of frontier AI systems. The law has since been modeled by other states. Newsom’s March 2026 executive order added civil rights and privacy protections to California’s procurement of AI technology and expanded the state’s own adoption of AI tools for government services. Additional legislation has addressed deepfake protections, AI watermarking requirements, performer digital likeness protections, and AI-generated robocall scams.
On the workforce investment side, the state has supported more than 674,000 earn-and-learn training opportunities since Newsom took office, including over 250,000 registered apprenticeships. The University of California’s Extension program serves more than 300,000 people annually. A new Degree Plus pilot combines a UC bachelor’s degree with skills-based certificates and paid internships. And the state has invested nearly $750,000 in the California Workforce Association to develop a statewide AI workforce strategy that will help local workforce boards prepare workers for emerging roles.
What the Tracker Cannot Do
The tool measures unemployment insurance claims — a dataset that captures workers who have lost jobs and filed for benefits. It does not capture workers who were never laid off but whose roles were restructured, whose hours were reduced, or whose tasks were automated while they remained employed in a diminished capacity. It also does not measure the jobs AI creates — the new roles in AI development, deployment, maintenance, and oversight that are growing alongside the displacement risk.
These limitations are inherent to any claims-based tracking system, and the California Policy Lab has been transparent about them. The tracker is an early warning instrument, not a comprehensive census of AI’s labor market effects. Its value lies in detecting shifts that warrant deeper investigation, not in providing definitive answers about whether AI is a net creator or destroyer of employment.
For a state that is home to 33 of the world’s top 50 private AI companies and simultaneously home to millions of workers whose occupations face meaningful automation exposure, the tension between those two realities is not abstract. The AI-Unemployment Tracker is California’s attempt to monitor that tension in real time — and to act on what it finds before the data becomes a crisis rather than a signal.
