A former OpenAI and Anthropic researcher has resigned from the artificial intelligence industry and issued a dire warning that the race to develop superhuman AI is moving forward without adequate safeguards.

Jacob Cookson, who worked at two of the world’s leading AI companies, announced that he had resigned from Anthropic after concluding that neither company was acting with sufficient responsibility.

“Neither company is behaving responsibly,” Cookson wrote, accusing the firms of “gambling with our lives” as they compete to develop superhuman systems capable of improving themselves.

Cookson warned that advanced AI systems could soon achieve capabilities far beyond those of humans, conduct sophisticated cyberattacks, dramatically accelerate scientific research and exert significant influence in the real world.

“Do not underestimate the power of this technology,” he wrote, adding that progress is not slowing.

Cookson claimed that individuals directly involved in developing the world’s most advanced AI systems genuinely believe the technology could “kill us all by the end of the decade.”

“This is not a marketing stunt,” he said, claiming that senior executives and researchers have privately expressed even more alarming concerns than those they acknowledge publicly.

According to Cookson, some OpenAI employees still do not fully grasp the potential “implications for civilization.” He said Anthropic is more aware of the risks but remains locked in the race out of fear that its competitors will reach superhuman AI first.

Cookson described the situation as an “arrogant gamble” and called on leading AI companies to reach agreements that would allow them to slow development. He said a temporary pause in improving AI models should be considered until researchers can better guarantee that the systems will remain under human control.

Anthropic safety researcher Evan Hubinger publicly backed Cookson’s warning, writing, “Jacob is right.”

Hubinger said he believes there is a greater than 10% chance that AI could cause the deaths of all humans within the next decade. He warned that despite major investments in safety, researchers still have no proven method of ensuring that a system more capable than humans will continue operating according to its intended goals.

Cookson concluded by urging AI employees to consider what will happen when they are asked to activate a superhuman system whose behavior and potential risks they do not fully understand.

(YWN World Headquarters – NYC)