The workers gaining the most from Claude fear it the most

The common playbook on AI safety at work has always been simple: become a power user so you don't get left behind. If you are the person orchestrating the tools, your job is safe.

Anthropic just blew that theory out of the water.

Anthropic released an economic follow-up study tying its Claude usage metrics to feedback from 80,508 workers across the globe. The headline result is a stark paradox: the very people extracting the largest productivity gains from AI are the ones losing the most sleep over losing their jobs.

Power users are running scared

According to the survey, workers whose day-to-day duties lean hardest on Claude voiced displacement fears at three times the rate of those who barely touch the tool.

Leading the panic are software engineers. That makes intuitive sense. When you spend all day watching an AI autocomplete your logic, fix obscure edge cases, and scaffold entire systems in seconds, you develop a front-row seat to your own potential obsolescence. Those who understand the model best know exactly how fast the runway is shrinking.

Right behind them are early-career professionals, who voiced the loudest displacement fears of all. This directly reinforces earlier signals from Anthropic pointing to a noticeable hiring slowdown for recent graduates in the U.S. When the entry-level stepping stones get automated away, junior workers are left wondering how they will ever climb the ladder.

Faster output just buys you more work

Meanwhile, the day-to-day payoff from AI is turning into a double-edged sword.

Most respondents confirmed that Claude delivers genuine speedups, knocking out tasks faster and handing back pockets of free time. But that free time does not stay empty for long.

Instead of a lighter workload, workers reported an expanded scope of responsibilities and a heavier pile of expectations. When management sees a task that once took two days wrap up in two hours, the bar immediately resets. The productivity numbers go up, but worker sentiment around AI has cratered, leaving employees trapped in a treadmill of their own efficiency.

Why it matters

This study shatters the naive tech optimism that leaning into AI is a guaranteed personal hedge. For individual developers and knowledge workers, mastery over the latest model does not buy peace of mind anymore; it just raises your quota. If you are leaning on AI to do raw execution, you are training the very system that will eventually commoditize that output. The only real leverage moving forward is owning problem definition, critical judgment, and organizational accountability.

For the tech industry at large, the warning lights are blinking red. Cutting off junior hiring to juice short-term margins is an act of corporate seed-corn eating. Without juniors cutting their teeth on routine tasks, companies will soon face a severe shortage of senior talent who actually understand how the machinery works under the hood. Anthropic deserves credit for releasing unflinching data about its own user base, but looking at the findings, there are virtually no easy solutions in sight to ease the growing friction.