All articles
AI & JobsJuly 4, 2026 · 8 min read

Will AI Take My Job? What Anthropic's Research Actually Shows

AI is not replacing workers at scale — not yet. According to Anthropic's March 2026 labor market report, there is no statistically significant increase in unemployment for workers in AI-exposed occupations since ChatGPT launched.

A
Anastasiia Veselova
July 4, 2026

The metric everyone's been missing: "Observed Exposure"

Most AI-and-jobs research asks: what can AI theoretically do? Anthropic introduced a different question: what is AI actually doing, in real workplaces, right now?

They call it observed exposure — a measure that combines what AI is theoretically capable of with real usage data from Claude conversations in professional settings.

33%
Actual vs. theoretical coverageIn Computer & Math roles, AI could theoretically assist with 94% of tasks. In practice, observed exposure sits at 33%. The gap between capability and deployment is enormous.

Infrastructure, processes, budget, organizational readiness — all of it creates meaningful lag. The central insight: theoretical capability ≠ actual displacement.

Who's most exposed — and what's happening to them?

The three most exposed occupations: Computer Programmers (75% task coverage), Customer Service Representatives, and Data Entry Keyers (67%). These jobs have real AI usage, not just theoretical potential.

And yet: unemployment among the most exposed workers has not increased. The trend lines for high-exposure and low-exposure workers look nearly identical since late 2022.

14%
Hiring of young workersHiring of workers aged 22–25 in exposed occupations has slowed by approximately 14%. Not layoffs — fewer open doors for people entering the field.

What this means if you're job searching right now

The jobs aren't disappearing overnight. But the bar for getting through the door has quietly risen. Recruiters have fewer open roles in high-exposure fields — more competition for each position. Your resume and application have to work harder.

Not because AI took the job — because fewer jobs are being posted.

whynotapply.me

When every application counts more, send the right one

A generic resume sent to 30 companies is the wrong strategy. See exactly how your experience maps to a specific job description — and what to fix.

Try now
FAQ

Common questions

Not at scale. Anthropic's 2026 research found no measurable increase in unemployment for workers in AI-exposed occupations. Automation is happening at the task level within roles, not through mass layoffs.
Computer programmers (75% task coverage), customer service reps, and data entry workers show the highest observed exposure. But even in these roles, actual automation remains far below what's theoretically possible.
Observed exposure is a metric introduced by Anthropic that measures AI's actual usage in professional settings — not theoretical capability. It combines task data from the O*NET database with real Claude conversation data.
Hiring in AI-exposed fields has slowed, especially for workers aged 22–25. Fewer job openings means more competition for each role. The problem isn't that you lost a job — it's that there are fewer to apply to.
Focus on application quality over volume. Tailoring your resume and cover letter to each specific role significantly improves response rates — especially in fields where AI is compressing hiring.
All articles