Axios, Fortune, Reuters, and Business Insider contributed to this report.
The biggest names in AI are starting to sound like they live in different futures.
On one side, Anthropic is still warning that artificial intelligence could wipe out huge chunks of white-collar work. On the other, OpenAI is backing away from the doomier script and saying the job fallout has been much smaller than it expected. The result is a noisy, confusing mess for companies, regulators and workers trying to figure out what comes next.
Anthropic co-founder Chris Olah said at a Vatican AI ethics conference that there is “a real possibility” AI could displace human labor on a massive scale. That lines up with the more cautionary tone from Anthropic CEO Dario Amodei, who has repeatedly argued that AI could upend office work.
Sam Altman, meanwhile, is sounding a lot less alarmed. The OpenAI chief told Commonwealth Bank of Australia CEO Matt Comyn that he had been “pretty wrong” about how fast AI would erase entry-level white-collar jobs. He said he thought the impact would already be bigger than it is.
“I’m delighted to be wrong about this,” Altman said, adding that AI has not yet caused the kind of jobs apocalypse he once feared.
The shift is striking because Altman was one of the loudest voices warning about AI-driven labor shock. A year ago, he was talking about entire categories of jobs disappearing. Now he says the human side of work still matters too much for AI to fully take over, especially in jobs built around judgment, tone and actual back-and-forth with people.
He even tested that theory himself. Altman said he had tried letting AI handle Slack and email responses, then started taking some of that work back. The experiment, he said, made him rethink how much of employment depends on human interaction that does not translate neatly into automation.
Amodei has also softened his tone somewhat. He once warned that AI could wipe out half of white-collar jobs. More recently, he has framed the technology less as a destroyer and more as a productivity machine. In his newer view, AI may automate most of a job, but that does not necessarily mean the job vanishes; it may just change shape and expand.
Goldman Sachs CEO David Solomon has been pushing back on the panic for a while, arguing that history says economies usually create new work after major disruptions. His point is simple enough: the economy has absorbed huge technological shifts before, and it probably will again.
The data so far backs neither extreme very cleanly.
Yes, tech layoffs have piled up. Meta has cut thousands of workers, and other companies including Coinbase, Block, Pinterest and Shopify have tied restructuring to AI or AI-related cost pressures. But there are also signs the labor market is bending, not breaking. Stanford researchers say recent unemployment increases have mostly hit sectors with the least AI exposure. Indeed reports that software engineering job openings are up year over year even as overall openings fall. LinkedIn’s chief economist says AI has helped generate around 1.3 million new job postings.
At the same time, some companies are realizing the payoff from AI is messier and more expensive than the sales pitch suggested. Uber has said AI costs are getting harder to justify. Microsoft is reportedly scaling back some Claude Code licenses as the bill for AI tools climbs.
That is the part that keeps getting missed in the loudest forecasts. AI may cut jobs in some places, add them in others, and reshape a lot more than either camp wants to admit. The clean story is tempting. The real one looks uneven, expensive and pretty confusing.
Altman and Amodei may still disagree on the scale of the threat, but they have both landed in a more cautious spot than before. The jobs apocalypse, at least for now, has not arrived. And the smartest answer may be that the damage and the gains will show up at the same time, just not in the same places.
The latest news in your social feeds
Subscribe to our social media platforms to stay tuned