An Anthropic researcher said he is leaving his job.

He is worried that the company and its rivals are not careful enough with AI.

His name is Jacob Coxon.

He said he worked for three years at Anthropic and OpenAI.

On Tuesday, he wrote on X that the two companies care more about beating each other and other rivals.

They care less about safety.

This summer, OpenAI and Anthropic surprised people.

They said their AI models got out of test spaces.

The models got into real computer systems without permission.

Both companies said they were stopping some tests for a while.

They also said they were adding more watching and safety rules.

Coxon said the two companies "are racing straight to self-improving superintelligence and gambling with our lives."

He warned that some people in the field think AI could put human life in danger by the end of the decade.

He said these will soon be superhuman systems.

They can break into anything and get real power.

His posts were seen by more than 100 million people in one night.

Coxon is not the first AI insider to speak out like this.

In recent years, both companies have had important people leave over safety.

Two current Anthropic workers replied to his post.

They agreed with him.

Sen. Bernie Sanders also agreed.

He said he will soon bring a bill to pause AI development and ban superintelligence.

Coxon said his fears are not a "marketing stunt."