In June 2025, OpenAI CEO Sam Altman warned that entire job categories could vanish as AI advanced. AnthropicCEO Dario Amodei went further, claiming roughly half of all white-collar positions faced existential risk from automation, with unemployment potentially hitting 10 to 20 percent within several years. Those predictions circulated widely, dominated policy discussions, and landed in the mental calculus of workers across every sector trying to figure out whether their careers had a future. This week, both men said they were wrong. Altman, in an interview with Commonwealth Bank CEO Matt Comyn, said he was “pretty wrong” about AI’s economic impact and that the displacement he feared simply hasn’t materialized. Amodei has reframed automation not as a job destroyer but as a productivity multiplier. Both OpenAI and Anthropic are preparing IPOs targeting late 2026, with valuations near or above $1 trillion and $380 billion respectively.
For the better part of a year, two of the most powerful figures in technology used their platforms to describe a labor market on the edge of structural collapse. That narrative had consequences that did not reverse themselves when the CEOs changed their minds. A recent workforce survey found that 24 percent of workers say AI is already negatively affecting their mental health, placing it on par with job demands and financial concerns as a primary stressor. 48 percentof employees say their job has negatively impacted their mental health over the past year, and 84 percent report burnout is affecting their productivity. Workers did not manufacture that anxiety. They were handed it by the people now walking it back ahead of a roadshow.
In 2025, companies directly attributed approximately 55,000 job cuts to AI — twelve times the number reported in 2023. Some economists have argued that companies used AI as a convenient cover for routine layoffs, attaching a technological rationale to workforce reductions that had more to do with margin management than automation. The apocalypse narrative made that cover easier to reach for. When the most prominent voices in AI are publicly forecasting mass displacement, any company that wants to cut headcount has a ready-made explanation that sounds structural rather than strategic. Independent research complicated the job-apocalypse narrative throughout — Yale Budget Lab found no significant shifts in occupational mix or unemployment for high-AI-exposure jobs since ChatGPT‘s 2022 launch. That data was available while Altman and Amodei were making their predictions. Yann LeCun, one of the founding figures of modern AI research, spent the same period telling anyone who would listen to ignore the noise and consult economists instead. The economists were saying something different than the CEOs. The CEOs had larger platforms.
The reversal itself raises a credibility problem that neither CEO has addressed. If the original predictions were made in good faith based on genuine expert analysis, the question is why two of the most well-resourced technology companies in the world — with access to labor economists, policy researchers, and their own product data — got it so dramatically wrong. If the predictions were made strategically, to shape regulatory conversations, establish AI as an inevitable force, or position their companies as navigating a crisis only they could solve, then the correction is not a sign of intellectual honesty. It is a brand management decision arriving on an IPO timeline. Tech layoffs citing AI as a driver exceeded 115,000 through May 2026 while both men were publicly forecasting mass displacement. Workers, policymakers, and the broader public organized their understanding of the economy around those forecasts. The same institutional credibility that amplified the original warning is now being deployed to walk it back — and there is no particular reason to trust the new version more than the old one, except that this version is better for the stock price.
The structural question SSC contributor Bryson Davis raised in his May 23 analysis of the April jobs report applies here with particular force. That report found 650,000 Black men had exited the labor force since November 2025 — not counted as unemployed because they had stopped looking, not captured in the headline figure that called the economy stabilized. The sectors most affected — logistics, warehousing, transportation — were not collapsing because of AI automation. They were collapsing because of tariff policy, fuel costs, and trade disruption. But in an information environment saturated with AI displacement narratives, the specific structural causes of specific communities’ specific economic pain became harder to see clearly. The dominant story absorbed everything else.
That is the less-discussed cost of elite-driven narrative. When the people with the most powerful platforms decide what the economic story is — even when the data does not fully support it — the correction, when it comes, does not reach the workers who restructured their lives around the original claim. It reaches investors preparing to price an IPO. Altman said he took criticism for fearmongering but believed it was necessary to address what he perceived as genuine risks. The workers who spent 2025 in therapy offices discussing the end of their careers, or who accepted worse terms because they felt replaceable, or who watched their employers cite AI as justification for layoffs that had nothing to do with automation — they did not get that caveat in real time. The prediction was wrong. The damage it did was not hypothetical. And the correction arrived on a schedule that serves the people who made the mistake, not the people who lived inside it.
Related SSC coverage: “The Numbers the Headline Isn’t Telling You” (Bryson Davis, May 23) · socialstorytellerscollective.substack.com