Anthropic released a report Thursday claiming its Claude AI agents now log more work hours than human employees, that entry-level technical hires dropped from 30% to 9% since March 2025, and that engineers spend 22% of their time managing AI agents. The company frames itself as a preview of the economy's future: fewer junior workers, more agent-wrangling, single-person startups commanding billions. It's a vivid picture. It's also entirely self-reported by a company preparing for a stock market debut. The timing matters more than the content. Anthropic could go public as soon as November, but its financial story is ugly: an $80bn net loss in the first half of 2026, a junk credit rating from June, and a valuation that investors have already marked down from $2tn to $1.5tn. Publishing a report that says 'our product is so good it replaced our own workers' is not labor economics — it's a product demonstration dressed as research. Yale social scientist Julián Posada, author of Platform Extractivism, called it out directly: 'It is written from the perspective of a company that wants to sell shares.' The report's methodological gaps are significant. No independent auditor verified the claims. No worker sentiment data was collected. The figures show hiring shifts and chat-log summaries, but nothing about output quality, error rates, or whether the remaining engineers are burning out managing what one unnamed Anthropic employee described as 'an army of Claudes, with every one asking for my attention.' Dylan Baker of the Distributed AI Research Institute noted the obvious: 'I wouldn't think really thoughtful, well-designed products are coming out of that team.' Meanwhile, the broader claim that AI is driving job losses faces its own evidence problem. Morgan Frank at Case Western Reserve published a paper showing AI-exposed job markets deteriorated well before ChatGPT launched in November 2022. State unemployment claims reveal the decline started in early 2022. Frank pointed to pandemic-era hiring corrections and 'AI-washing' — bosses crediting AI for layoffs they'd have made regardless — as likelier explanations. This fits a pattern. Dario Amodei predicted last year that AI would eliminate half of entry-level white-collar jobs and write 90% of code within six months. Neither happened. The predictions served the same function as this report: positioning Anthropic's product as inevitably transformative, which is exactly what you want investors to believe when you're about to sell shares. The report's proposed remedy — stock ownership for workers displaced by AI — is revealing. It concedes the harm ('it is difficult to imagine pro-worker AI adoption without financial stakes') while proposing a solution that requires workers to bet their compensation on the company's stock price. For a firm with an $80bn net loss and a junk credit rating, that's not worker empowerment. It's risk transfer. Anthropic's cloud computing partner CoreWeave is drowning in AI debt with its stock down 25%, Pew data shows a majority of Americans are increasingly wary of AI and job displacement, and the company did not respond to the Guardian's request for comment. The report lands in a moment where the AI industry needs its narrative of inevitability more than ever — precisely because the financial fundamentals are beginning to crack.
