Anthropic just ran the numbers on its own AI's capabilities, and the white-collar jobs most at risk aren't the ones you'd expect.
The Summary
- Anthropic's head economist shared data showing AI can already handle far more work than most companies realize
- The jobs most exposed aren't low-skill, they're middle-tier knowledge work where tasks are repeatable and rules-based
- The economics chief discussed which specific roles could be eliminated, not just automated
The Signal
Anthropic hired an economist, which tells you everything about where this is headed. Not philosophy, economics. Peter McCrory, Anthropic's economics chief, is looking at labor markets the way climate scientists look at ice cores. He's measuring exposure, not possibility.
The data coming from inside Anthropic is different from the usual think-tank speculation. These are numbers based on what their AI can actually do right now, not what might happen in five years. The surprise: high-exposure jobs aren't call center workers or data entry clerks. Those jobs either already got automated or are too chaotic for current AI to handle reliably. The bullseye is middle-management knowledge work. Compliance officers. Junior financial analysts. Legal research associates. Mid-tier HR roles. Jobs where the work follows patterns, decisions follow rules, and outputs can be checked against known-good examples.
This matches what companies are quietly doing. They're not firing people yet, they're just not backfilling positions. A legal team that used to hire two research associates every year now hires one and gives them an AI copilot. Revenue per employee goes up. Headcount stays flat or drifts down. No headlines, no layoffs, just slow replacement through attrition.
The Implication
If you're in a role where you spend most of your day applying known rules to new situations, you're in the blast radius. The move isn't to panic, it's to climb the ladder toward judgment calls AI can't make, or sideways into work that requires physical presence, deep client relationships, or creative synthesis. The jobs getting hollowed out are the ones companies used to hire new grads into, which means the entry ramp to white-collar work is disappearing. If you're hiring or managing people, the question isn't whether to use AI, it's how fast you can retrain your team to work at a higher altitude.
Sources: Fortune Tech | Fortune Tech