The Retraction
On May 26, 2026, Sam Altman stood in front of a Commonwealth Bank of Australia conference in Sydney and said something unusual: he admitted he had been wrong.
“I’m delighted to be wrong about this,” he told CBA Chief Executive Matt Comyn. “I thought there would have been more impact on entry-level white-collar jobs being eliminated by now than has actually happened.”
Altman said OpenAI had been “roughly right” on its technological predictions since launching ChatGPT in 2022, but “pretty wrong on the social and economic impact.” He added: “I don’t think the jobs apocalypse is coming.”
He even walked back his own AI use. He had experimented with letting an AI respond to emails on his behalf — the messages were labeled “this is Sam’s AI” — before reverting to writing some himself. His takeaway: “It was an amazing example to me of we really do care about people.”
What He Was Walking Back
Altman helped build the expectations he is now revising. Over multiple years, he and other AI leaders warned that AI would soon be capable enough to displace significant portions of the knowledge workforce — particularly entry-level roles requiring reading, writing, coding, and routine analysis.
Those warnings were treated seriously not because they were fringe views, but because they came from the people building the systems. Boards took them seriously. Policy makers took them seriously. A generation of young workers entering white-collar professions took them seriously.
When the person who helped create the fear says “I was wrong,” that is legitimately newsworthy.
What the Data Actually Shows
Altman’s headline is accurate at one level: there has been no sudden, broad “apocalypse” of white-collar job losses that reshapes the unemployment rate in an obvious way. The overall U.S. unemployment rate has not spiked. The economy continues to add jobs. Most office workers still have jobs.
But the picture is meaningfully more complicated below that headline:
Entry-level hiring has collapsed. Entry-level job postings have declined approximately 35% since January 2023 across white-collar sectors. Employers who once hired junior staff to do research, summarize documents, draft emails, or write boilerplate code are doing that work with AI assistants instead — not hiring the entry-level worker at all.
Young software developers are being hit hardest. Among 22–25-year-olds in AI-exposed roles, employment fell roughly 16% from late 2022 to mid-2025. For young software developers specifically, the decline was nearly 20% relative to the late-2022 peak. These are not marginal numbers.
AI-attributed tech layoffs are real. In the first six months of 2025 alone, an estimated 78,000 tech jobs were attributed to AI-driven restructuring. These are mostly mid-sized positions — not the entry-level apocalypse and not the senior-leadership apocalypse, but the middle rungs of career ladders that used to bridge junior and senior work.
The gender dimension is striking. 79% of employed U.S. women work in high-automation-risk roles, compared to 58% of men — reflecting that clerical, administrative, and customer service positions, which are being automated fastest, remain disproportionately held by women.
The Distinction That Matters
Altman’s prediction was for broad white-collar displacement: that even senior knowledge workers would lose jobs. That has not happened. Senior analysts, senior developers, senior managers — their employment has been largely stable or even grown, because AI tools make experienced workers more productive rather than replacing them.
What has happened — and what the data confirms — is a specific, structural shift at the bottom of the career ladder:
- Fewer junior hires
- Fewer entry-level postings
- Narrower paths from “just graduated” to “established in a career”
This is not an apocalypse. But it is also not nothing. It is a structural change in how careers form, concentrated on workers who are the least established and least able to adapt quickly.
Altman’s original fear was roughly correct about the mechanism — AI absorbs routine, repeatable tasks — but wrong about the layer of the workforce that would feel it first. He expected senior and mid-level workers to be disrupted fastest. Instead, the disruption started at the entry point and is working its way up more slowly than anticipated.
What This Means for AI Practitioners
If you are building AI products or deploying AI in an organization, Altman’s Sydney remarks contain a signal worth taking seriously:
AI’s economic impact is real but non-linear. The systems are capable enough to eliminate junior workloads at scale. They are not yet capable enough to eliminate senior judgment at scale. Products built on the assumption that AI will fully replace senior knowledge workers are probably two to four years early.
Entry-level talent pipelines are thinning. If your organization relies on growing its own senior talent from junior hires, AI is quietly shrinking your pipeline. The industry is reducing the number of junior developers and analysts who will become senior developers and analysts in five years.
The politics of AI and jobs will intensify. Altman can credibly say the “apocalypse” did not arrive on schedule. But the workers who graduated in 2023 or 2024 into an entry-level market that declined 35% have a different lived experience. As that cohort ages and earns political weight, AI companies should expect their employment narrative to face sustained scrutiny.
The Honest Summary
Sam Altman made fear about AI job losses, admitted on May 26 that the fear was larger than the facts warranted, and expressed genuine relief about being wrong. That is a better-than-average response from a technology executive to a mistaken prediction.
The employment data broadly supports his revised view: no broad white-collar apocalypse, no mass unemployment crisis, no sudden reshaping of the workforce in the way he once implied might happen.
But “not an apocalypse” and “no meaningful disruption” are not the same statement. For entry-level workers, especially in software and knowledge roles, the disruption is already here — it just arrived quietly, through thinning job postings rather than visible layoffs.
Altman says he now “understands more about why it hasn’t” gone further. That explanation — and whether it holds — will be the more important story to watch.
This article is based on publicly reported conference remarks from May 26, 2026, and publicly available employment research data through mid-2026. ChatForest research does not include hands-on employment market testing.