For years the debate about AI and jobs has run on prediction rather than proof. That is starting to change. As The Straits Times reports, drawing on government figures compiled by Bloomberg, payrolls in the US information and financial-activities sectors, the two industries that adopted AI fastest, are now falling by an average of 28,000 a month in 2026. The striking part is the contrast: the wider labour market added more than 113,000 jobs a month through May. The economy is growing while a specific, AI-heavy slice of it quietly shrinks.
The corroborating numbers point the same way. Challenger, Gray & Christmas, which tracks announced layoffs, has attributed almost 102,000 job cuts to AI so far this year, and says the tech sector alone accounts for a third of all layoffs announced in 2026. "It's certainly making an impact as we speak in a way that no technology has before," said the firm's chief executive, John Challenger, who expects finance to be the next sector hit hardest. His reasoning is structural: office and administrative roles such as customer service reps, bank tellers, and claims processors make up about a quarter of financial-industry employment, a larger share than in any other major sector, and those are exactly the tasks AI automates most readily.
There is even a research pattern beneath the anecdotes. A study from Stanford's Digital Economy Lab found that employment has weakened in occupations where AI automates the work, while holding up where AI merely assists the worker. A California Policy Lab tracker found the state's highest concentration of unemployment claims from AI-exposed workers in finance and insurance, with information and professional services close behind. The effects, the researchers wrote, "may be starting to surface." Public anxiety has already arrived: a June Pew poll found 40 percent of Americans expect AI's impact on society to be largely negative, against just 16 percent who expect it to be positive.
And yet the economists urging caution have a strong case too, which is what keeps this honest. At the macro level it remains too early to isolate AI's signature from everything else moving through the economy. Ryan Nunn of the Yale Budget Lab notes that layoff data for finance shows no unusual spike, suggesting AI may be working first through slower hiring and attrition rather than mass firings, a quieter mechanism that thins the ranks without a headline. Barclays economist Pooja Sriram offers a sharper alternative reading: much of what companies call AI-driven efficiency may really be old-fashioned cost-cutting, dressed up to justify the enormous sums already spent on the technology. In that account AI is less the cause of the cuts than the alibi for them.
Both things can be true. AI can be genuinely displacing some tasks while also serving as convenient cover for reductions firms wanted anyway. For the people caught in it, the distinction barely matters. Bill Matonte, a software engineer, landed a job at Citigroup in six weeks back in early 2025; laid off this April, he has now spent six months interviewing without an offer. "It's really stressful," he said. That is the texture the data will never quite capture: not a sudden apocalypse, but a labour market that is slowly getting harder to re-enter in precisely the corners where the machines are learning fastest.