Photo by Mahmudul Hasan on Unsplash
53 percent. That is how far California software development job postings fell from their late-2022 peak — and for the first time, a state government now has a dedicated tool to determine whether artificial intelligence is actually the reason.
According to Google News, California's Governor's office confirmed on June 25, 2026 that the state deployed the California AI-Unemployment Tracker (CAIT), a first-in-the-nation public dashboard that cross-references unemployment insurance claims against measured AI exposure levels by occupation. No other state has attempted anything at this scale or with this methodology.
What the Tracker Actually Measures
CAIT draws from California Employment Development Department unemployment insurance records spanning January 2017 through May 2026, deliberately excluding the pandemic window (March 2020 through January 2022) to isolate AI-related trends from pandemic-era noise. The California Policy Lab at UCLA built the underlying methodology, classifying occupations into three tiers using two separate AI-exposure measures.
The first is potential exposure: occupations where large language models could reduce task completion time by 50% or more, scored at ≥0.49. The second is observed exposure: actual usage frequency recorded on Anthropic's Claude platform, scored at ≥0.107. Occupations landing in the top 25% by either measure are classified as high-exposure. That dual measurement is what separates CAIT from prior academic estimates — using Claude usage data means it captures where AI tools are already embedded in real workflows, not merely theoretically capable of replacing tasks.
The tracker updates monthly and breaks results down by 14 regional planning units, industry, age, education, race, ethnicity, and gender. That granularity matters enormously, because the statewide picture and the regional picture are telling very different stories.
The Numbers Worth Paying Attention To
At the state level, the headline finding is cautiously optimistic. As of May 2026, CAIT shows no statewide surge in unemployment claims for AI-exposed occupations compared to pre-pandemic baselines. Dr. Ben Hyman, California Policy Lab Senior Researcher, said directly: "Right now, we are not seeing evidence of large-scale AI-related layoffs in California's labor market. But we do see patterns in certain regions like the Bay Area, in certain tech-heavy sectors, and among highly AI-exposed workers with college degrees."
That second half of his statement is where the story actually lives. The San Francisco Bay Area recorded what researchers describe as a "sharp and sustained increase" in unemployment claims among workers in high-AI-exposure occupations. Among software developers specifically, employment for workers aged 22 to 25 fell nearly 20% from a late-2022 peak. Software development job postings dropped 53% over the same period. Federal Reserve data shows US programmer job growth fell 50% following ChatGPT's launch. Separately, CAIT's data shows college-educated workers in high-AI-exposure occupations began filing elevated unemployment claims after ChatGPT-3.5's November 2022 release, and those claims remained elevated through May 2026. Across all AI-exposed fields, employment for workers aged 22 to 25 declined 13% since 2022.
That is a tight cluster of converging signals, not coincidence.
Chart: Measurable decline in tech employment and hiring across multiple indicators since the November 2022 ChatGPT launch, as of May 2026.
It is also worth noting that California's own state government acknowledged using six high-risk AI systems in government operations in 2026, after reporting zero such systems as recently as 2025 — including the COMPAS recidivism-scoring tool used in criminal justice decisions for over a decade. That admission suggests AI deployment has outpaced institutional awareness of it, which is precisely the gap CAIT is designed to close on the labor market side.
Why This Matters Beyond California
California's move did not happen in a policy vacuum. Governor Newsom signed an executive order in May 2026 directing the state to prepare workers, small businesses, and communities for AI-driven economic disruption, with CAIT serving as the primary evidence-gathering infrastructure. At the federal level, Missouri Senator Josh Hawley introduced bipartisan legislation requiring companies to disclose AI-related layoffs to the federal government. New York lawmakers proposed an "AI Dividend" program tied to worker displacement. Vermont Senator Bernie Sanders issued formal warnings about AI displacement threatening American workers broadly.
Meanwhile, corporate data is moving in one direction. An HR Dive survey found that as of June 2026, nearly 37% of companies expect to have replaced some jobs with AI by year-end, with 3 in 10 already having done so. Anthropic CEO Dario Amodei has predicted that AI could eliminate half of all entry-level white-collar positions within five years — a claim that lands differently now that Anthropic's own platform usage data forms one of CAIT's two core exposure metrics.
This connects to a larger capital story that AI Trends recently examined: the $2.59 trillion flowing into AI infrastructure is accelerating the very disruption CAIT is now attempting to measure. The money flows toward the technology; the displacement surfaces in unemployment insurance claims. CAIT is the first government instrument trying to close that loop in real time.
Till von Wachter, UCLA Economics Professor and California Policy Lab Faculty Director, framed the tracker's purpose plainly: "AI is advancing quickly, and workers' concerns about what that could mean for their jobs are real. This new tracker helps replace speculation with evidence."
Where Your Leverage Actually Lives
Here is the practical read for anyone in an AI-exposed occupation — or managing personal finance in an economy where that exposure is expanding faster than most people are tracking.
Check your occupation's exposure tier. The California Policy Lab's methodology is publicly accessible through capolicylab.org. If your role qualifies as high-exposure (top 25% by either the ≥0.49 potential score or the ≥0.107 observed Claude usage score), the Bay Area data suggests displacement pressure is already active in tech-heavy markets — not approaching, but present. Knowing your tier is the starting point for any honest financial planning conversation about career risk.
The age and education pattern is the real signal here. CAIT's most specific finding is that college-educated workers in high-AI-exposure occupations began experiencing elevated unemployment claims after November 2022, and that pattern persisted for more than three years through May 2026. Entry-level workers in AI-exposed fields (ages 22 to 25) saw employment fall 13% since 2022; software developers in that same cohort dropped nearly 20%. If you're early-career in a high-exposure field, the standard personal finance guidance of three to six months of emergency savings was calibrated for cyclical downturns. Structural displacement events have longer, harder recovery tails. Adjust accordingly.
The career conversation script, if you need one: "I've been tracking industry data on AI task automation. I want to understand which parts of my current role the company sees as AI-assisted over the next 12 to 18 months — and where you'd want me to develop deeper skill." That is not a defensive question. It is the question of someone who reads the evidence and frames the conversation before a restructuring announcement forces it. The market doesn't reward people who avoid that conversation; it tends to reward the ones who initiate it.
Frequently Asked Questions
How does California's AI unemployment tracker actually work?
CAIT cross-references California Employment Development Department unemployment insurance claims against AI-exposure scores for each occupation. It uses two metrics: a potential exposure score (≥0.49, where AI tools could reduce task completion time by 50% or more) and an observed exposure score (≥0.107, based on actual usage frequency on Anthropic's Claude platform). Data covers January 2017 through May 2026, excluding the pandemic period. As of June 26, 2026, results are publicly accessible through the California Policy Lab at UCLA, broken down by region, industry, age group, education level, and demographics, with monthly updates going forward.
Is AI taking jobs in California right now — what does the data actually show?
As of May 2026, the statewide picture shows no large-scale surge in AI-related unemployment claims, according to CAIT. However, the San Francisco Bay Area recorded a sharp and sustained increase in unemployment claims among high-AI-exposure workers. Employment for software developers aged 22 to 25 fell nearly 20% from a late-2022 peak, and software development job postings declined 53% over the same period. The summary: no statewide crisis visible yet, but measurable and sustained displacement concentrated in tech-heavy regions and among early-career workers in high-exposure occupations.
What jobs are most at risk from artificial intelligence in the near term?
Based on CAIT's methodology, high-risk occupations are those scoring in the top 25% by either potential AI exposure (≥0.49) or observed Claude platform usage (≥0.107). Current evidence from CAIT points to college-educated, early-career workers in software development and other tech-adjacent roles as experiencing the most acute near-term displacement pressure. Nearly 37% of companies surveyed by HR Dive expect to have replaced some positions with AI by end of 2026, with 3 in 10 already having done so. Anthropic's CEO has predicted that half of entry-level white-collar jobs could be eliminated within five years — and Anthropic's own platform usage data now forms part of CAIT's exposure measurement methodology.
Bottom line: In my analysis, CAIT is less significant for what it found than for what it establishes — a government acknowledging, with actual unemployment insurance data, that AI-driven displacement is already measurable in specific pockets even as the aggregate looks calm. The Bay Area numbers are not statistical noise. They are, in all likelihood, a preview of what broader AI adoption will surface in other high-exposure markets over the next two to three years. If your financial planning doesn't account for the possibility of a longer-than-typical job search window in an AI-exposed field, June 2026 is a reasonable moment to recalibrate that assumption.
Disclaimer: This article is for informational purposes only and does not constitute financial, legal, or career advice. Editorial commentary is based on publicly reported information and does not represent independent verification of the underlying data. Research based on publicly available sources current as of June 26, 2026.