How AI Is Quietly Eliminating 10,000 White-Collar Jobs a Month — Without a Single Announcement

When desks are no longer being refilled, a certain silence descends upon an office floor. Not overly dramatic, with security guards and bankers’ boxes—nothing so dramatic. Just a retirement letter here, a resignation letter there, and an unapproved requisition that would have taken their place in the inbox of some vice president. The chair remains in place. The monitor remains in place. The position does not return. In early 2026, that quiet has turned into a kind of ambient hum in American corporate offices, and it’s growing louder.

When you put the numbers together from disparate sources, you get a picture that official statistics haven’t quite caught up to. Challenger, Gray & Christmas, a company that has been monitoring layoffs since 1989, estimated that in 2025, AI was directly responsible for about 55,000 job losses. A rounding error in an economy with 125 million nonfarm jobs seems doable.

Key Information: AI and White-Collar Employment

However, modeling estimates from several research groups indicate that the real number of jobs lost or never created due to AI tools last year was closer to 200,000 or 300,000. One of the defining characteristics of this era is the discrepancy between what businesses claim to do and what they really do. According to a Duke University survey of 750 chief financial officers, which was released in collaboration with the Federal Reserve Banks of Atlanta and Richmond, roughly 44% of businesses intend to make some AI-related layoffs in 2026, which equates to about 502,000 positions in the overall economy. The researchers pointed out that white-collar jobs would account for half of those.

It’s difficult to ignore how cautiously businesses steer clear of the term “layoff” when AI is involved. Instead, it’s more subtle: two positions are discreetly incorporated into automation workflows, and a team of twelve becomes a team of eight after two individuals depart. Income increases. The headcount remains constant. No announcement. Not a headline.

The Labor Statistics Bureau Job openings in professional and business services are close to their lowest point in almost ten years, according to JOLTS data. This trend started in late 2022 and has only gotten worse. Similar findings can be found in LinkedIn workforce reports: tech postings decreased by over 30% at different times during 2023 and 2024, and this decline has spread to positions in finance, law, marketing, and administration. These figures do not indicate a recession. The economy is not shrinking. Businesses are just learning how to expand without hiring more staff.

Something that had been simmering for months came to a head in February with the Block episode. Jack Dorsey, who co-founded Twitter before it was taken over by someone else, declared that his fintech company was laying off more than 4,000 workers, or about 40% of its workforce, citing AI tools as the cause. The stock increased by 24%. It turned out that investors adored the concept of a business accomplishing more with drastically less.

Block was doing well, according to Dorsey, who also noted that gross profit was increasing. The issue, if you can call it that, was that fewer people were needed to complete the task. However, a few of those same workers were discreetly rehired within weeks. A design engineer was informed that his termination was the result of a “clerical error.” After their managers battled to get them back, others got calls. It felt like a business that used a machete when it needed a scalpel, making cuts first, asking questions afterwards, and then bandaging the injuries it didn’t mean to cause.

However, Dorsey’s prediction came to pass: most businesses would come to the same conclusion and implement comparable structural changes within a year, he wrote. Even though his approach was unusually direct, there is a feeling that he might not be incorrect after watching corporate earnings calls since then. Citing its agentic AI product, Salesforce eliminated 4,000 customer service positions.

Leaner structures are made possible by AI, according to Amazon, which eliminated 14,000 corporate positions. Through a hiring freeze driven by AI, Klarna cut 40% of its workforce, and by 2030, it plans to cut another third. Duolingo declared that it would no longer employ human contractors for any tasks that AI is capable of performing. These businesses are all doing well. The majority are reporting revenue that is either at or close to records. Survival is not the goal of the cuts. They have to do with margins.

The fact that this contraction doesn’t resemble earlier automation waves is what makes it so confusing. Physical, visible labor was replaced by factory robots. The machines were visible. On the assembly line, you could count the bodies that are missing. This time, the displacement takes place within email threads, project management software, and spreadsheets.

A language model absorbs the work of a junior analyst. Software that does not bill on an hourly basis completes a paralegal’s research tasks in a matter of seconds. Because the remaining writers use AI to draft twice as quickly, a content team shrinks from six to three. According to Anthropic’s own Economic Index, which was released in early 2025 and was based on actual usage data from its Claude model, the majority of AI adoption is concentrated in occupations that pay more than the median wage, such as software developers, financial analysts, writers, and legal researchers. Not the CEO. Not the surgeon. the center.

According to a Stanford study, since ChatGPT’s launch in late 2022, employment among early-career workers in AI-exposed occupations has decreased by 16%. 16 percent. That statistic is especially painful for a generation that was told a college degree would protect them from economic disruption. The rungs of the ladder that young professionals were expected to climb included entry-level writing, junior data analysis, tier-one customer support, basic financial modeling, and administrative coordination.

The rungs are getting thinner. Dario Amodei, CEO of Anthropic, issued a warning last year that AI might eliminate half of all entry-level white-collar jobs in five years, possibly increasing unemployment to ten or twenty percent. Mustafa Suleyman, Microsoft’s AI chief, provided a 12- to 18-month timeline for the widespread automation of professional tasks. It was referred to by Senator Bernie Sanders as a “economic earthquake.” Jerome Powell, the chair of the Federal Reserve, admitted that AI seems to be subtly changing the job market.

One aspect of all of this that merits greater attention than it receives is the phenomenon that some researchers refer to as “AI washing.” According to a Built In study, nearly 60% of hiring managers in the United States intend to make layoffs in 2026, with artificial intelligence being the most frequently mentioned cause.

However, only 9% of respondents claimed that AI had completely replaced some jobs. According to 60% of respondents, they highlight AI’s contribution to cost reductions because it sounds more appealing than stating that “we need to reduce costs.” It’s possible—even likely—that a large portion of what is referred to as AI displacement is actually traditional restructuring dressed in Silicon Valley attire. The technology offers protection. Shareholders nod in agreement. Instead of giving the decision a sense of choice, it makes it seem inevitable.

However, it would be incorrect to completely ignore the trend. The information about job openings is accurate. There are actual hiring freezes. It is not a statistical illusion that businesses are increasing revenue while maintaining or decreasing headcount. This is known as “headcount containment,” according to Gartner and others, using AI to handle growing workloads instead of hiring more people to handle them.

It doesn’t appear as jobless. It manifests as a career that never starts, a job that is never advertised, or a promotion that is perpetually postponed because the team above you no longer needs another member. Robert Solow, an economist, once joked that the computer age was evident everywhere but in productivity figures. Similar findings were made by the Duke CFO survey: businesses recognize AI’s potential but lack the corresponding financial outcomes. Executives are waiting for the returns to catch up while investing in the promise and making cuts based on expectations. It’s genuinely unclear if they do, and if the human cost in the interim is quantified at all. What has occurred can be counted by the BLS. The job that just never happened is more difficult to count.