news.today
Hello · Register today | LoginHello · My Account[—] [32]
Tech

OpenAI and fired safety researchers clash over dismissal reasons

Summarised from 5 outlets · Updated 9 Oct, 10:56 · Archive

OpenAI has fired three safety researchers, saying they breached trust by mishandling sensitive information; the researchers dispute this, claiming they were dismissed for raising safety concerns.

OpenAI dismissed three safety researchers last week after finding they had mishandled sensitive company information outside established procedures. The company said Jasmine Wang, Tomek Korbak and Mikita Balesni had violated policies on accessing and handling sensitive information and shared it with a third-party AI safety organisation. OpenAI characterised the breach as a significant violation of trust.

The three researchers have published an open letter disputing OpenAI's account of their dismissals. They claim they were fired for prioritising safety over the company's near-term interests. In the letter addressed to OpenAI's Safety and Security Committee and other advisory bodies, they state they were dismissed for raising concerns about the company's ability to monitor artificial intelligence systems.

The researchers say their communications before dismissal included calls for the company to pause developing models that would decrease the ability to monitor AI systems. They argue that raising such safety concerns and drawing on outside expertise were previously integral to working at OpenAI and were encouraged by the company.

OpenAI has stood firm on its decision. In a statement on Friday, the company said the dismissals were "not about raising safety concerns or speaking out". A spokesperson told the BBC that its investigation "confirmed that these individuals mishandled sensitive information outside established company procedures". The company added that the internal investigation uncovered breaches beyond what was outlined in the researchers' letter but did not provide details.

The researchers have expressed concern that the public manner of OpenAI's communications around the dismissals may have a chilling effect on colleagues. They wrote that staff have become "afraid to speak and operate in ways that, until last week, were an integral part of working at OpenAI". One researcher, Tomek Korbak, stated in a social media post that he had been raising safety concerns about the company's ability to monitor what AI agents think, describing this as one of the best tools for catching when they misbehave.

The dismissals occur amid growing debate about AI safety following several high-profile security incidents involving AI systems, including a breach at Hugging Face attributed to OpenAI. Multiple researchers at AI companies have called for the technology to be developed more slowly and for greater safeguards against risks posed by advanced AI systems.

OpenAI and fired safety researchers clash over dismissal reasons
Image: BBC World

Comments (0)

How it is being reported

In this story: OpenAI · Jasmine Wang · Tomek Korbak · Mikita Balesni · Hugging Face

This summary was written by AI from the headlines and standfirsts above, and states as fact only what at least two outlets report. How we use AI · Report a problem