OpenAI stands by firing of three AI safety researchers

OpenAI said Friday it stands by last week’s firing of three safety researchers, who it says violated its policies on handling sensitive information. The researchers say they were fired for raising safety concerns.

OpenAI says its investigation uncovered “a significant breach of trust beyond what’s outlined in the letter they published,” the BBC and CBS News reported. It has not given details, The Verge reported.

The researchers are Jasmine Wang, Mikita Balesni and Tomek Korbak. They say they were, until recently, safety or alignment employees at the company, the BBC reported.

What OpenAI says

OpenAI answered in a note from research leaders posted on X on Friday morning, the BBC and CBS News reported.

“Last week we parted ways with Jasmine, Mikita, and Tomek after a thorough investigation found they violated clear policies on handling sensitive information,” the company wrote, as carried by CBS News.

“We want to be very clear that these decisions were not about raising safety concerns or speaking out,” the note said, according to the BBC. OpenAI said it tolerates “good-faith mistakes” and does not fire workers for “raising concerns,” CBS News reported.

An OpenAI spokesperson told the BBC on Oct. 2 that the investigation “confirmed that these individuals mishandled sensitive information outside established company procedures.”

What the researchers say

The three published an open letter Thursday. In it and in social media posts, they said they had “acted in line with OpenAI’s mission and within the working norms of the time,” The Verge reported.

“I believe we were fired for prioritising safety over the near-term interests of OpenAI as a corporation,” Balesni posted on X, the BBC reported.

Korbak wrote that for months he had been raising safety concerns “that we’re losing the ability to monitor what AI agents think, one of our best tools for catching when they misbehave,” according to the BBC.

The researchers wrote that they are concerned communications around their firing have made former colleagues “afraid to speak and operate in ways that, until last week, were an integral part of working at OpenAI,” CBS News reported.

Where they agree

OpenAI backed one point in the letter. “We agree with the letter that preserving the monitorability of frontier models requires an industry-wide commitment, including from OpenAI,” it said, according to CBS News.

It also said it is finalizing contracts with third-party safety assessors and will announce details in the coming weeks, the BBC reported.

Outside safety work

Both sides describe work with outside safety groups as normal. The letter says staff “were encouraged to draw on the expertise of independent safety organizations,” CBS News reported. OpenAI said, “Many of our researchers already work with 3p safety organizations productively,” according to CBS News.

What’s next

Whether OpenAI details what it says went beyond the letter, and whether it announces details on the third-party assessors in the coming weeks, as it said it would.