A researcher who says they spent the past three years working on AI pretraining at both OpenAI and Anthropic has resigned from Anthropic, issuing a stark public warning about the direction in which leading artificial intelligence companies are moving.
In a post shared on X, the former Anthropic employee said they had left the company “today” after three years of pretraining research spanning Anthropic and OpenAI.
“I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic,” the post said.
The researcher then made a much broader criticism of the two AI companies, arguing that neither was behaving responsibly in the pursuit of increasingly capable artificial intelligence.
“Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives,” the researcher wrote.
The statement is striking not simply because of the warning itself, but because it comes from someone claiming direct research experience inside two of the companies most closely associated with advanced AI development.
A Rare Criticism From Someone Who Worked at Both Companies
What gives the resignation additional weight is the researcher’s stated professional history.
According to the post, the individual spent the last three years conducting pretraining research at both OpenAI and Anthropic.
Pretraining sits at the heart of the development process behind modern large AI models. In the researcher’s own account, that experience has now led to a conclusion that the competitive push among major AI developers is moving in an irresponsible direction.
The researcher did not single out Anthropic alone despite resigning from the company. OpenAI was also named directly in the criticism.
That makes the post less a conventional resignation announcement and more a warning about the wider trajectory of frontier AI development.
‘Racing Straight to Self-Improving Superintelligence’
The most consequential phrase in the post is the allegation that the companies are “racing straight to self-improving superintelligence.”
The researcher did not provide technical details in the statement supplied here about what they mean by self-improving superintelligence, how close such systems might be, or what specific company decisions prompted the resignation.
For that reason, those questions should remain open rather than being filled with speculation.
What is clear from the post is the researcher’s position: they believe the direction of development at both companies creates risks serious enough to publicly criticize after leaving Anthropic.
The phrase “gambling with our lives” also communicates the severity of that concern without establishing that a catastrophic outcome is certain.
Why the Resignation Could Draw Attention
Public debate around advanced AI often revolves around two competing pressures: building more capable systems and determining how those systems should be controlled.
This resignation enters that discussion from an unusual position.
Rather than coming from an outside critic, policymaker or commentator, the warning comes from someone who says they were directly involved in pretraining research at both OpenAI and Anthropic.
That does not by itself prove the researcher’s predictions or conclusions. It does, however, make the criticism notable because it reflects the personal assessment of someone describing first-hand experience inside two major AI developers.
For now, the most accurate reading of the statement is also the least sensational one.
A former Anthropic researcher has resigned and accused Anthropic and OpenAI of irresponsibly racing toward self-improving superintelligence, warning that the competition is “gambling with our lives.”
That is already a significant claim. Turning it into “AI is going to kill us all” would go beyond what the researcher actually said.
Also Read: Jimmy Kimmel Live Future: Why His 24-Year Late-Night Run Is Suddenly Under Question
