Anthropic asks job applicants not to use AI assistants when submitting résumés

In February 2025, Anthropic, the company behind Claude, one of the best AI assistants available, sent job applicants an unexpected request: don't use AI when applying. The request covers more than engineers. It applies to everyone else too: marketers, finance specialists, communications staff, and salespeople. At first glance, it looks like a paradox. On a second look, it's a diagnosis of a disease that has spread across the entire job market.

Why Anthropic is making this request

Anthropic didn't explain its decision in detail, but the point is easy to grasp: the company wants to see how candidates write on their own, without help from AI assistants. Every employer makes this kind of request. Show what you yourself can do, that these are your thoughts and your writing style, not neural network output styled to sound like you.

The problem is that the requirement has a fundamental flaw: it's nearly impossible to verify. How can anyone reliably tell a résumé a person wrote from scratch apart from one they sketched as a draft, ran through Claude, and then edited a few sentences of? There's no reliable technical way to check. That makes Anthropic's request a polite letter covering a sadder truth: the company has stopped trusting résumés as a source of information about candidates.

When the hiring mechanism broke

For years, the résumé worked as a reliable signal. A prospective employer looked at the text, the project descriptions, the wording, the grammar, and all of those details added up to a picture of how well the candidate could structure ideas, describe achievements, and write persuasively.

That mechanism is now broken:

• A strong candidate can quickly draft a résumé in Claude, polish it, and the employer won't know the difference
• A weak candidate can submit a near-perfect résumé written entirely by AI, and nobody will catch it
• Roughly half of candidates use AI when applying and half don't, but nobody knows who is who, and there's no way to check

As a result, the résumé has lost its informational value. It's just polished text of unknown origin. Anthropic's request is an attempt to restore an honest comparison between candidates, though it only works if everyone agrees to follow it.

The problem affects every major company

Anthropic has named the problem explicitly, but Google, Meta, OpenAI, Stripe, and AWS all face it too. Each of them sees the share of suspiciously flawless résumés grow year after year.

Some companies have chosen to dig deeper: instead of relying on résumés, they look at answers to tough interview questions, run coding challenges, and require candidates to work through case studies in real time. It's more expensive and slower, but it works, because an AI assistant helps less under those conditions.

Anthropic took a more honest route: it simply asked candidates not to use AI. The outcome will be the same either way, since people will either follow the request or they won't, but at least the issue is out in the open.

What it means

This isn't a solution. It's a diagnosis. Anthropic is showing that once a technology becomes universally available, the old ways of validation turn into shadow theater. A résumé can look great and still be both flawless and completely useless as a source of information.

In the long run, the hiring market will adapt. New signals will emerge: maybe real-time video interviews, maybe quick tests with open access to Claude. Anthropic's request is only the first step toward making hiring mean something again.

*Meta is designated an extremist organization and banned in Russia.