California launches a real-time tracker to measure whether AI is actually killing jobs
California Governor Gavin Newsom has unveiled a public data tool designed to answer a question that has been mostly speculation until now: is AI actually costing people their jobs?
The system, called the California AI-Unemployment Tracker (CAIT), cross-references unemployment insurance claims with detailed occupational data to identify which workers are most exposed to AI displacement. Each occupation gets scored on how susceptible it is to automation, and those scores are mapped against real claims data month by month. The tracker updates monthly and functions as an early-warning system — more radar than final diagnosis.
The first report, covering data through May 2026, offers a mixed picture. At the aggregate level, there is no evidence of an AI-driven surge in unemployment claims across California’s massive labor force. But the numbers look different when you zoom in.
Since ChatGPT 3.5 launched in late 2022, unemployment claims have risen steadily among college-educated workers in occupations classified as highly susceptible to AI replacement. That trend is most pronounced in the San Francisco Bay Area — the AI industry’s own backyard.
By sector, two categories show the clearest signals.
| Sector | Trend Since ChatGPT 3.5 (Late 2022) | Current Status (May 2026) |
|---|---|---|
| Professional Services | Sharp spike soon after launch, then gradual decline | Still elevated above pre-AI baseline |
| Information Technology | Gradual increase through 2024, then decline through 2025 | Returned to baseline by late 2025 |
The state is careful not to over-interpret these patterns. The tracker does not prove that AI caused any specific layoff. Macroeconomic factors — interest rates, consumer demand, post-pandemic restructuring — all play a role. Instead, the system is designed to give policymakers lead time. If certain occupations start showing sustained increases in claims, the state can route affected workers toward job counseling, retraining programs, skills upgrading, and health insurance support before the problem becomes a crisis.
US media coverage has focused on the tracker’s broader significance. Until now, the debate about AI and jobs has lived mostly in the hypothetical. CAIT moves it toward measurable, longitudinal data. Researchers can track, month by month and occupation by occupation, which parts of the labor market are actually changing — and how fast.
What makes the tracker noteworthy is not what it proves today, but what it makes possible over time. Policy decisions around education funding, unemployment benefits reform, and workforce development have long been made with limited visibility into how technology shifts the labor market. A system like CAIT, running for years, could replace intuition with evidence.
The tracker is still young. One report does not make a trend. But it is the first government-run monitoring system of its kind in the United States, and it marks a shift from guessing about AI’s impact on labor to watching it happen in real time.