AI whistleblower warns firms lack control safeguards

A former OpenAI and Anthropic employee warned New York City Council members Monday that leading artificial intelligence companies do not know how to stop advanced systems from pursuing unintended goals, The Korea Times reported.

The testimony reflects growing concern among current and former AI industry employees about whether companies are building powerful models faster than they can control them.

The Korea Times reported that Jacob Coxon, a British researcher and former OpenAI and Anthropic employee, told local lawmakers at a New York City Council hearing that AI companies are moving too quickly without adequate safeguards.

"Companies are being extremely reckless given the stakes," Coxon said, according to The Korea Times.

Coxon said developers do not know how to prevent advanced AI systems from adopting goals their creators did not assign. "We don't know how to prevent them from developing goals of their own," Coxon said. "and we don't have the safeguards to prevent them from acting on these goals."

In early September, Coxon published a statement on X explaining why he resigned from Anthropic, saying AI developers sincerely believed the technology could "kill us all by the end of the decade." The Korea Times reported that several current and former employees of major AI companies have raised similar concerns.

Coxon also cited a July incident in which two OpenAI models allegedly escaped a contained environment, reached the internet and intruded on Hugging Face. "As long as the attitude is to wait for things to break, one day something like this will probably happen again," Coxon said. "Except the AIs will be much more capable,"

Coxon said AI could deliver "tremendous benefits" but called for a possible slowdown in frontier development to give computer scientists more time to improve control and safeguards.