AI-linked job cuts hit 117,000 in 2026, and now there's a dollar figure

The Center for AI Safety counts just over 117,000 US job cuts attributed to AI since January 2026 — and it puts the price tag at $8.5 billion in estimated annual wage losses. The San Francisco-based nonprofit's Wage Loss Dashboard, updated this month with September data, converts AI-attributed layoffs into dollars and cents for every state. That matters because the national argument over AI and employment has mostly run on anecdotes, and this is an attempt to replace those with numbers anyone can check.

What does the dashboard actually measure?

The tool shows a map of layoffs since January 2026, plus a drop-down that lets you view the whole country or a single state. It multiplies the number of AI-attributed job cuts by the average annual salary in that state to estimate lost wages, then divides by population to get an average loss per person.

Nationally, that works out to $25 per person per year. The state numbers diverge sharply:

• California: $56 per person, on the back of just over 27,000 AI-attributed layoffs
• Alaska: $14 per person, with 138 job losses

Those aren't direct observations. As CAIS explains: "The dashboard estimates AI-attributed cuts for each state, converts them to estimated annual wage loss, then divides by state population to estimate the average loss per person. State estimates are not observed AI layoffs. We apply each month's national share of cuts citing AI to that state's announced job cuts." In plain terms: the state figures are modeled, not counted one by one.

The dashboard also translates the per-person figure into everyday expenses, like a tank of gas — a deliberate effort to make abstract layoff statistics feel concrete. The takeaway: even modest-sounding per-person numbers hide concentrations that hit some states much harder than others.

Where does the data come from?

Three sources feed the model. Challenger, Gray, and Christmas supplies monthly reports on layoff announcements by state, plus national job cuts that companies themselves credited to AI. The Bureau of Labor Statistics provides each state's annual mean wage. US Census Bureau population estimates complete the calculation.

The base numbers are substantial on their own. In 2025, Challenger, Gray, and Christmas identified over 54,000 job cuts that companies publicly linked to AI — roughly one in 20 layoffs that year. CAIS argues that counting layoffs alone is more abstract than showing what they mean in money. "There has been significant news coverage of job loss from AI, but it can be hard to quantify how these cuts affect the average person's finances," CAIS wrote of the project. "This dashboard aims to help illuminate the monetary costs of AI on American families and communities."

Why does the dollar framing matter?

CAIS's argument is economic, not just statistical. Layoffs ripple outward through local economies — grocery stores, small businesses, rent. "One person's layoff means less income for every business they support. Multiply that effect across thousands of workers in multiple industries, and you start to see the real economic weight of AI displacement," the researchers wrote. "In addition, AI-attributed job cuts aren't spread evenly. States with large tech and media sectors are experiencing greater impact."

That unevenness is the policy stake. A state like California absorbs both the tech sector's AI adoption and its job cuts, which is one reason its per-person loss is more than double the national average. The point, in short: displacement costs land locally, even when the technology is national.

How reliable are 'AI-attributed' layoffs?

Here the picture gets fuzzier, and CAIS is upfront about it. The dashboard doesn't include "jobs created by AI, productivity gains, re-employment, or secondary economic effects." That's partly because experts still disagree on whether AI will stimulate the economy or outpace job displacement.

The layoffs themselves are also murky. The data counts job cuts that companies said were tied to AI — but there's no single definition of what that means, and without full transparency from companies, it isn't always provable that a given layoff resulted directly from AI replacement.

A parallel tracker, Jobloss.ai, run by the Alliance for Secure AI, faces the same problem from the other direction. It only logs layoffs of at least 100 jobs at once, accompanied by some mention of AI or proxy terms like "automation" and "productivity." The standard is intentionally loose. "We don't require an executive to say 'we replaced a team with AI.' This is a moderate standard, deliberately: it captures the broad pattern of AI-justified workforce reductions while excluding clearly unrelated cuts," the Alliance for Secure AI wrote of its methodology.

CAIS says the dashboard isn't meant to shame any company for adopting AI. "AI is a technology being adopted by companies for many reasons," CAIS wrote. "Our goal is to make the costs of that adoption visible, so workers, communities, and policymakers can respond with real information instead of guesses." Bottom line: the numbers are estimates built on self-reported corporate claims, and readers should treat them as a pattern indicator rather than an audit.

How fast is AI taking over remote work?

The dashboard's second section tracks capability, not just cuts. Using the Remote Labor Index (RLI) benchmark, which CAIS co-developed to test AI agents on real freelance tasks like web development and video production, it measures how quickly agents are improving.

"When RLI launched, the best AI agent completed just 2.5% of projects at professional quality. Eight months later, that number has risen to 15.8%," the researchers wrote. That near-16% automation rate is doubling every 3.7 months. Claude Fable 5 currently leads the benchmark, followed by OpenAI's GPT-5.5.

Capability isn't the same as adoption — RLI measures what models can do, not the freelance projects they're actually winning from humans. But CAIS flags the combination that worries it: AI-attributed job cuts rising at the same time as agents get measurably more capable of the work being cut. The signal to watch: that doubling rate, which turns a modest 16% into something much larger within a year if it holds.

What are lawmakers doing about it?

Legislation can't stop companies from choosing AI systems over human hires in most cases. But several states, including California, have recently passed laws aimed at ensuring firing and other disciplinary decisions aren't fully automated by AI systems. That's a narrow guardrail — it addresses process, not displacement itself.

The gap between what regulators can reach and what the dashboard measures is the story's open question. As the RLI numbers climb and the layoff counts accumulate, CAIS is betting that visible, localized cost data will give workers and policymakers something firmer than guesswork to respond with.