Anthropic Is Measuring AI-Induced Job Loss Risk : These 10 Occupations Are Most Vulnerable
Anthropic has launched a new “observed exposure” metric to track how AI could affect jobs. Its analysis shows roles such as programmers, data entry workers and financial analysts are among the most exposed to automation, though large-scale AI-driven job losses have not yet appeared.
Artificial-intelligence developer Anthropic has introduced a new analytical tool to measure the risk of job displacement caused by AI, identifying several white-collar professions that may be particularly vulnerable as generative AI adoption accelerates.
The company’s research introduces a metric called “observed exposure,” designed to assess how AI could affect employment by combining the theoretical capabilities of large language models with real-world workplace usage data from systems such as Anthropic’s AI assistant Claude.
Researchers say the goal is to create an early-warning indicator that can track potential labour-market disruptions before large-scale job losses become visible in official employment statistics.
Occupations most exposed to AI
According to the analysis, the following 10 occupations currently show the highest exposure to AI-driven automation based on how many of their tasks could be performed by AI tools:
- Computer programmers
- Customer service representatives
- Data entry keyers
- Medical record specialists
- Market research analysts and marketing specialists
- Sales representatives in wholesale and manufacturing (excluding technical products)
- Financial and investment analysts
- Software quality assurance analysts and testers
- Information security analysts
- Computer user support specialists
In some cases, a large share of job tasks could theoretically be handled by AI. For example, about three-quarters of programming tasks could potentially be performed by AI systems, the study suggests.
Early signals but limited job losses so far
Despite high levels of theoretical exposure in certain occupations, Anthropic’s analysis found no clear evidence of widespread AI-driven unemployment so far.
However, the research does point to early labour-market shifts. Hiring among younger workers particularly those aged 22 to 25 appears to have slowed in some AI-exposed professions, indicating that automation could affect entry-level opportunities first.
White-collar roles face greater risk
The findings suggest that AI may disrupt knowledge-based and administrative occupations more than jobs that require physical labour. Roles such as construction workers, mechanics or cooks remain relatively insulated because current AI systems cannot perform tasks that require physical presence or manual work.
The gap between what AI systems could theoretically automate and how they are actually used today also remains significant. Many tasks that AI could perform are not yet widely adopted in professional environments due to legal constraints, technical limitations and the need for human oversight.
Monitoring AI’s labour-market impact
Anthropic researchers say the “observed exposure” index is intended to help economists and policymakers track how rapidly AI adoption could reshape employment patterns.
While the current labour-market impact appears limited, the company warned that the gap between AI’s potential capabilities and its present use suggests future disruption could grow as adoption increases.
The research comes amid intensifying debate over how generative AI systems may transform the global workforce, particularly in sectors where work involves writing, analysis and information processing tasks that modern AI models increasingly perform.