California's AI Unemployment Tracker Is Live for WDBs
California Executive Order N-6-26 gives the state 90 days to build its own AI Workforce Playbook to represent each region’s labor market and economy. You can learn more about the order in our post, "California AI Workforce Playbook: What Every WDB Director Should Know Before October 15, 2026." On June 25, 2026, Governor Newsom's office launched the California AI-Unemployment Tracker (CAIT), built in partnership with the Employment Development Department (EDD) and the California Policy Lab (CPL) at the University of California Los Angeles.
A question workforce development board leaders can ask today is this:
Can you name your region's top five AI-exposed occupations?
CAIT is the first public tool that starts to answer that question. However, it has limits. Understand them before your board exclusively leans on it for anything official.
What Is the California AI-Unemployment Tracker?
CAIT is not an EDD product built in-house. It's built by the California Policy Lab, a nonpartisan research center at UCLA, using EDD's own unemployment insurance data. EDD's own documentation notes that CPL's findings don't necessarily reflect the views of the state's Labor Market Information Division.
That distinction matters. CAIT is an independent analytical layer sitting on top of state data. It's not an official state finding about what's causing job loss in any region. Rather, this tool provides early signs of possible AI-induced job loss.
Why Does Executive Order N-6-26 Require This Dashboard?
The order is written to measure first and regulate later. It creates no new obligations for private employers. It creates no private right of action. Multiple employment law firms tracking the order since May confirm this reading.
What the order does is direct state agencies to study AI's labor market impact and report back. CAIT is the EDD AI dashboard for watching that impact unfold nearly in real time, using unemployment claims as the signal. It sits alongside, and separate from, the provision directing EDD and local boards to build the regional AI workforce playbook due October 15.
CAIT answers one question: what's happening statewide.
The playbook mandate asks a different question: what's happening in your region? And crucially, what is your board doing about it?
A second, later deadline sits inside the same order. Within 180 days of the May 21 signing, so by mid-November 2026, the Labor and Workforce Development Agency must review the state's WARN Act and recommend whether it needs amending to address AI-driven layoffs specifically. That review hasn't produced anything public yet.
CAIT is the first deliverable to land.
The WARN Act recommendation is the next one to watch.
How Does the Tracker Actually Measure AI's Impact on Jobs?
CAIT assigns an occupation an AI exposure score. It draws on unemployment insurance claims, records of workers laid off through no fault of their own from a covered California employer who applied for benefits.
When someone files a claim, they self-report their occupation from a standardized list. CAIT assigns that occupation two separate exposure scores. Potential AI Exposure measures whether AI models can cut the time needed for an occupation's tasks by at least half, a standard developed by OpenAI and academic researchers.
Observed AI Exposure measures how often an occupation's tasks actually show up in real usage of Anthropic's Claude, drawn from the Anthropic Economic Index. CPL uses both together, and reports the results as largely consistent across both..
CAIT breaks this data down further. It shows trends by county of residence, industry, age, gender, race and ethnicity, education, and region.
But the chain of measurement stays the same at every level: a claimant reports an occupation, the occupation gets an exposure score, the score gets tracked over time.
At no point does CAIT verify that AI caused any individual person's layoff.
It measures exposure by job category. It does not confirm causation for any specific job loss.
What Does the Data Show So Far?
Statewide, CAIT's own initial analysis says there is no evidence of a surge in unemployment claims tied to AI-exposed occupations. That's the headline finding from CPL's own researchers.
But the data isn't flat everywhere. Claims from college-educated workers in high-AI-exposure occupations increased after ChatGPT-3.5's public release in 2022. Those claims stayed elevated through May 2026. Workers in low-exposure occupations show no comparable shift over the same period, relative to pre-pandemic 2019 and early-2020 levels.
In the San Francisco Bay Area specifically, workers in high-AI-exposure occupations show a sharp, sustained increase in claims over this same period. This is the one region CPL names directly in its published findings.
Statewide, claims are also elevated in the Information and Professional Services sectors specifically.
CPL's analysis found no large disproportionate increases by race, ethnicity, gender, or age among high-exposure claimants.
Read this data as what it is: an early sign of possible AI-related disruption in specific groups and one named region. Not proof that AI caused any of it.
What Are CAIT's Limits for Local Workforce Planning?
CAIT supports a county-level view for every California workforce development board. But CPL's own published report only names one region directly, San Francisco. No public finding has been issued for most other California counties.
That is not a reason to wait. It is exactly why CAIT is a starting point and not a finished analysis. One detail worth knowing before your board opens the dashboard: CAIT reports regions using 14 statewide regional planning units (RPUs), not individual county names.
For example, a board in Riverside or Imperial needs to identify its RPU first. Then, it can read the dashboard's regional breakdown against it, pairing findings with its own local labor market data before drawing any conclusion.
Two more limits worth stating. Every number in CAIT is an exposure signal, not a causation finding. If a program officer asks how your board identified its exposed occupations, the honest answer is "CAIT flagged this occupation as high-exposure, and we cross-checked it against our own caseload."
Not "CAIT proved this."
Second, CAIT fulfills a different part of the order than the WDB playbook mandate does. Pulling numbers from CAIT does not, by itself, satisfy your board's October 15 obligation.
What Should Your Board Do With This Data Right Now?
Four steps. None require waiting for EDD to reach out.
Pull your county's numbers directly from CAIT. Don't wait for a state press release to confirm your region matters.
Cross-reference those numbers against your own program participant data. Which occupations are already showing up in your caseload right now?
Treat CAIT as a starting input for your regional occupation exposure analysis. It is not a substitute for building one.
Write down your methodology as you go. If a program officer later asks how your board identified its exposed occupations, "CAIT data plus our own regional caseload" is a defensible answer. A list built from CAIT alone is not.
CAIT is useful precisely because it's the first real, checkable tool this order has produced. Boards that open it this week and cross-reference it against their own data move closer to an October 15 playbook. Boards that wait for the next headline don't.
Frequently Asked Questions About About California's AI-Unemployment Tracker
Is CAIT the same thing as the October 15 AI workforce playbook mandate?
No. CAIT fulfills a separate provision of Executive Order N-6-26 directing EDD to build a public dashboard from unemployment insurance data. The playbook mandate, due October 15, 2026, requires EDD and local boards to jointly build a regional strategy identifying AI-exposed occupations and dislocated worker strategies. CAIT can inform that work. It doesn't complete it. Read the full text of Executive Order N-6-26 here.
Does CAIT prove AI is causing layoffs?
No. CAIT assigns an AI exposure score to a claimant's self-reported occupation, based on published research into which job tasks are most exposed to AI generally. It does not verify that AI caused any specific layoff. Read every number as an early signal worth investigating, not a confirmed cause.
Exposure varies significantly by region and industry mix. Nationally, occupations with high rates of routine cognitive tasks, including administrative support, data entry, claims processing, and customer service, face the highest near-term augmentation or displacement risk.
In California, regional labor markets vary widely. A board serving the Central Valley faces a different exposure profile than one serving Silicon Valley or Los Angeles. A compliant AI Workforce Transition Playbook requires a regional occupation exposure analysis built on local labor market data, not national averages.
Where can I access CAIT??
The California Policy Lab hosts CAIT directly, along with its full methodology and downloadable data. View CAIT and its full methodology here.
Does CAIT replace my board's own occupation exposure analysis?
No. CAIT gives your board a statewide, and increasingly county-level, reference point. Your board's regional occupation exposure analysis, one of the three deliverables under the October 15 mandate, still requires your own local labor market data and program knowledge. CAIT is an input, not a finished analysis.
Digna Vita AI builds the WDB AI readiness strategy your board brings to EDD's playbook. Schedule a call to talk through where your board stands today.
