Anthropic CEO says AI will take half of office jobs and increase unemployment by 10-20%

Taras Mishchenko - 29 May, 10:51 AM

Anthropic CEO Diego Amodei warns that artificial intelligence could take away half of all entry-level office jobs and increase unemployment to 10–20% within the next 5 years, Axios reports.

Ironically, Amodei’s comments come after his company unveiled its next-generation AI models, the Claude Opus 4 and Sonnet 4, which can work on tasks longer and perform them more accurately, potentially increasing the risk of what the Anthropic CEO warns about.

However, Amodei calls on the US government and AI companies to stop downplaying the threat: according to him, lawyers, consultants, financiers and IT beginners will face massive layoffs as AI agent systems increasingly perform routine tasks at or above human level. "Most of them are unaware that this is about to happen," he said, and suggested holding official briefings for lawmakers, requiring companies to disclose data on the use of AI models in different industries, and introducing a tax on AI income to fund employee retraining programs.

According to Amodei, AI has the potential to cure cancer, grow the economy, and balance the budget, but without active measures to preserve jobs and distribute the benefits, millions of people will find themselves out of work, and inequality will only deepen.

Big companies are already replacing humans with AI assistants: Mark Zuckerberg predicts that mid-level programmers will start losing their jobs this year. Zuckerberg told Joe Rogan’s podcast in January: "In 2025, AI systems at Meta and other companies will be capable of writing code like mid-level engineers." He said that this would eventually reduce the need for humans to do that job. Shortly after, Meta announced plans to cut its workforce by 5%. Amodei added that "once businesses see the benefit of AI replacing humans, they will do it en masse."

To prevent this scenario, Amodei suggests:

Publish data on AI usage by profession in real time;
Encourage the complementation of human labor, not its complete replacement;
Introduce a "token tax" on AI transactions to redistribute profits.

"You can't just step in front of the train and stop it. The only move that's going to work is steering the train — steer it 10 degrees in a different direction from where it was going. That can be done. That's possible, but we have to do it now," concludes the CEO of Anthropic.