AI not yet cutting jobs but slowing hiring in high-risk roles: Anthropic study

In the business and finance domain, about 94 per cent of job tasks could potentially be handled by AI, yet only 28 per cent are actually being done using AI tools. Roles such as computer programmers, financial analysts and customer service workers show the highest exposure to automation.

As the job market watches AI advance with unease, new research by Anthropic, the company behind Claude, finds that while AI has the potential to reshape jobs, early signs show lower actual adoption due to legal and technical constraints. The study also notes that AI is causing slower hiring and has not yet led to job destruction.

For instance, for a role in the Business and Finance domain, roughly 94 per cent of the job tasks could theoretically be performed by LLMs, while only 28 per cent of the tasks are already being done using AI, suggesting a long journey for AI to impact jobs.

Jobs most exposed

Computer programmers, financial analysts and customer service jobs were noted to have the highest risk of exposure. On the other hand, manual jobs like mechanics, lifeguards and bartenders are expected to be insulated.

Interestingly, the report also indicates that, despite rising automation, companies are not yet laying off in vulnerable roles; instead, they have stopped hiring for high-risk jobs, thereby impacting opportunities for fresh graduates.

“Using survey data from the US, we find no impact on unemployment rates for workers in the most exposed occupations, although there’s tentative evidence that hiring into those professions has slowed slightly for workers aged 22-25,” Anthropic said.

Who faces risk

A demographic analysis by the authors of the report also reveals that women, white people, older, and higher-paid individuals are more likely to be at risk.

Published on March 6, 2026