Concerns About The Pace Of AI Development Are Mounting. OpenAI Is Reportedly Considering Slowing Down.

OpenAI CEO Sam Altman reportedly said the company is considering slowing down the development of cutting-edge AI as concerns over the possibility that unregulated progress could get out of control mount.

Speaking in a company-wide meeting this week, Altman said OpenAI could make the decision, perhaps along with other industry peers, but some may not agree to do it, Bloomberg detailed.

The dangers of AI progressing too quickly and getting out of control dominated the conversation this week, particularly after Antrhopic researcher Jacob Coxon accused both OpenAI and Anthropic of moving too aggressively in their pursuit of increasingly capable systems.

"The people building AI earnestly believe that it could kill us all by the end of the decade," Coxon wrote in his post. "This is not a marketing stunt."

His concerns center partly on the competition among major AI developers and the difficulty of monitoring increasingly capable models. Coxon told Axios that he had not seen Anthropic compromise its safety practices while he worked there, but said competitive pressure could eventually lead companies to "cut corners" or "skip steps in the oversight process."

He also said AI models have become increasingly capable of recognizing when they are being evaluated. Systems knowing when they are being tested, something that once appeared far removed from current AI capabilities, is now a regular part of working with advanced models, Coxon said.

Coxon's warnings have received public support from other researchers at Anthropic. Evan Hubinger, who leads the company's alignment science work, responded to Coxon's resignation post by saying he agreed that AI could pose an existential threat and put the probability of AI killing all humans within the next decade at above 10%.

His resignation adds to a series of departures from major AI companies tied to concerns about safety. Researchers have previously left OpenAI and other AI labs after disagreements over how companies are balancing the development of more capable systems with work intended to keep those systems under control.

In this context, a Senate subcommittee began investigating OpenAI's handling of the high-profile hacking of Hugging Face in July.

Senator Josh Hawley, the chairman of the Senate's Subcommittee on Disaster Management, sent a letter to OpenAI CEO Sam Altman in which he describes "growing allegations of the existential risk of new AI products."

Hawley described OpenAI's handling of the incident as "reckless," adding that there is still information that could be unearthed. "The auditors were given complete transcripts of AI agent activity for only two days, when the events leading up to the Hugging Face breach took place over weeks," reads a passage of his letter.

"The American people deserve to know the details of what went on in the Hugging Face incident and other incidents of AI models going rogue. This investigation will seek those answers," Hawley concluded.