Companies Are Laying Off Workers for AI's Potential -- Not Its Performance
In March 2025, New York became the first U.S. state to require employers to disclose whether AI or automation contributed to mass layoffs. Since then, 162 companies have filed layoff notices affecting 28,300 workers.
The number that checked the AI box: zero.
This includes Amazon, which publicly warned that AI would reduce headcount across 30,000 roles, and Goldman Sachs, which internally linked reductions to AI productivity gains while topping New York's layoff charts with over 4,100 affected workers.
Companies are publicly attributing workforce reductions to artificial intelligence while privately declining to make the same claim under legal obligation. And the research increasingly suggests a reason: most of these layoffs have nothing to do with AI's actual capabilities.
The Anticipation Economy
A Harvard Business Review survey of 1,006 global executives, published in January 2026, found that AI-attributed layoffs are "almost completely in anticipation of AI's impact" rather than responses to proven AI performance. Only 2% of organizations reported large headcount reductions tied to actual AI implementation. The rest were cutting based on what they expect AI to do eventually.
That finding lands differently when paired with a National Bureau of Economic Research paper surveying nearly 6,000 CEOs and CFOs across the U.S., UK, Germany, and Australia. More than 90% of business managers reported that AI had no impact on their employment over the past three years. Eighty-nine percent said the same about labor productivity.
The layoffs are not a response to technological disruption. They are a bet on it.
The ROI Problem
If companies were cutting workers because AI had rendered their roles redundant, the technology would be delivering measurable returns. It is not.
An MIT report published in August 2025, based on 150 executive interviews, 350 employee surveys, and analysis of 300 public AI deployments, found that 95% of companies investing in generative AI are seeing zero return -- what researchers called the "GenAI Divide." U.S. businesses have collectively invested between $35 billion and $40 billion in AI initiatives. About 5% have achieved measurable results.
The pattern is consistent: companies announce AI-driven workforce reductions to signal innovation to investors, while the underlying technology remains incapable of absorbing the work those employees performed.
Even OpenAI's CEO sees it. Sam Altman told CNBC at the India AI Impact Summit in February 2026 that some companies are engaged in "AI washing" -- "blaming AI for layoffs that they would otherwise do." When the person most commercially invested in AI's success says the layoff narrative is overstated, the narrative deserves scrutiny.
The Regret Curve
The scrutiny is arriving in the form of data.
Forrester's research found that 55% of employers already regret AI-attributed workforce reductions. The firm predicts that more than half of these layoffs will be quietly reversed, with roles returning offshore or at reduced salaries.
Gartner's February 2026 prediction is more specific: by 2027, 50% of companies that attributed headcount reductions to AI will rehire staff to perform similar functions -- often under different job titles. Gartner's own October 2025 survey of 321 customer service leaders found that only 20% had actually reduced agent staffing because of AI. Most said headcount remained steady even as they handled more customers.
The most instructive case study is Klarna. In 2024, CEO Sebastian Siemiatkowski announced that AI chatbots were doing the work of 700 customer service representatives. By mid-2025, the company was rehiring human agents after customer satisfaction dropped and complaints about robotic, inflexible responses mounted.
"We focused too much on efficiency and cost. The result was lower quality, and that's not sustainable." -- Sebastian Siemiatkowski, Klarna CEO
Klarna's reversal is not an anomaly. It is a preview.
Build the Tool, Lose the Job
Perhaps the most disorienting pattern is the number of workers who built the systems cited as their replacements.
At King, the Candy Crush developer, level designers, UX researchers, and narrative writers spent months building and training internal AI tools. In July 2025, approximately 200 of them were laid off and replaced by those same tools. One developer noted: "Most of level design has been wiped, which is crazy since they've spent months building tools to craft levels quicker."
A 2026 Gartner survey found that 64% of organizations implementing AI had used existing employees to create training data, but only 22% had been transparent about how that data would affect future staffing decisions. A parallel SHRM survey found that 71% of workers involved in AI training programs did not believe their employer was being honest about long-term staffing implications.
Block provides the largest-scale example. In February 2026, CEO Jack Dorsey cut approximately 40% of the company's workforce -- roughly 4,000 employees -- attributing the decision to AI. Every remaining employee is now required to use generative AI tools daily, with compliance tracked. But analysts noted that Block's headcount had ballooned from 3,800 in 2019 to over 10,000, and the CBC reported that the cuts looked more like "a mix of AI efficiency gains and an overdue clean-up of corporate bloat."
Amazon's trajectory is similarly instructive. After CEO Andy Jassy warned that AI would mean the company would "need fewer people doing some of the jobs that are being done today," the company cut over 30,000 corporate roles across 2025 and early 2026. Then in March 2026, Amazon's retail website experienced four high-severity incidents in a single week -- including a six-hour outage -- after an engineer followed inaccurate guidance from an AI agent that had pulled from an outdated internal wiki.
The company is now introducing what it internally calls "controlled friction" for AI-assisted changes to critical systems. The humans, it turns out, were performing a function.
The Numbers in Context
According to Challenger, Gray & Christmas, AI was cited in approximately 55,000 U.S. layoffs in 2025. That is a twelve-fold increase from 2023. But it represents less than 5% of the 1.2 million total job cuts announced that year. More than twice as many were attributed to restructuring. Nearly six times as many were linked to government spending cuts.
The trend line matters more than the absolute number. AI-attributed layoffs grew from 4,600 in 2023 to 12,700 in 2024 to 55,000 in 2025. Early 2026 data suggests acceleration, with over 30,000 tech jobs already cut and AI agents specifically cited in roughly 9,200 of them.
But the composition of these cuts tells a different story than the headlines. The HBR survey found that the areas most affected -- entry-level roles, customer service, programming -- are exactly the areas where AI augmentation has shown the most promise but the least proven replacement capability. Companies are not eliminating roles where AI has demonstrated it can do the work. They are eliminating roles where AI might eventually do the work.
The Counter-Examples
Not every company is following this pattern, and the exceptions are instructive.
Walmart announced in February 2026 that it would provide free AI training to all 1.6 million U.S. employees through a partnership with Google, with no planned headcount reductions. "We'll have roughly the same number of people we have today," CEO John Furner said.
The distinction is strategic, not philosophical. The HBR survey found that 57% of executives expect AI to increase headcount over the next year, compared to only 15% who expect it to decrease headcount. The majority view is that AI changes the nature of work, not the volume of it. The layoffs are coming from the minority -- but they are louder.
What This Means for Job Seekers
The practical implications are counterintuitive.
The job market is harder, but not primarily because of AI. Applications per opening have surged, time-to-hire has stretched to an average of 42 days, and the "low-hire, low-fire" pattern that defined 2025 is continuing into 2026. AI is a contributing factor in some sectors, but economic uncertainty, cost-cutting, and post-pandemic workforce corrections explain far more of the contraction.
The anxiety is outpacing the reality. Employee concerns about AI-related job loss jumped from 28% in 2024 to 40% in 2026. Nearly half of Gen Z job seekers believe AI has already diminished the value of their college education. But the NBER data shows that 90% of executives report no actual AI impact on employment -- yet. The fear is doing real damage to career decisions and mental health independent of any technological displacement.
The jobs being eliminated are not disappearing permanently. Gartner's prediction that half of AI-attributed cuts will be reversed by 2027 suggests that many of these roles will return -- likely rebranded, possibly at lower compensation, but functionally similar. Workers who maintain their skills and stay in the market will be positioned for that correction.
Throughput still matters in a competitive market. With success rates per application hovering between 0.1% and 2%, consistent output is not optional. The difficulty is that maintaining both volume and quality across dozens to hundreds of applications is a mechanical problem as much as a strategic one.
The gap between AI's narrative power and its operational reality is the defining feature of the current labor market. Companies are making real workforce decisions based on speculative capabilities, and more than half are already regretting it. For job seekers, the market is difficult, but the difficulty is not permanent, and much of it has less to do with artificial intelligence than the headlines suggest.
The best response to a harder market is not paralysis -- it is velocity.