Former OpenAI researcher says he was fired after raising security concerns to board
Former OpenAI employee Leopold Aschenbrenner has revealed that he was fired from the company after raising security concerns about AI to the company's board of directors.
Leopold Aschenbrenner - China/US Super Intelligence Race, 2027 AGI, & The Return of History
OpenAI employee says he was fired for raising security concerns to board
OpenAI Whistleblower Claims He Was Dismissed For Raising Security Concerns - WinBuzzer
https://winbuzzer.com/2024/06/05/openai-faces-allegations-of-retaliation-from-former-employee-xcxwbn/
Aschenbrenner graduated from Columbia University as valedictorian at the age of 19 in 2021 and joined OpenAI's security team ' SuperAlignment ' as an early member in 2023. He worked with the team's co-founders Ilya Satkivá and Jan Reich, but was fired from the company in early 2024. Since then, he has been working as a security researcher. Satkivá and Reich retired in May 2024, and the SuperAlignment team wasdisbanded .
Speaking on AI expert Dwarkesh Patel's podcast, Aschenbrenner explained that he was fired 'for bringing security concerns about OpenAI to the board of directors.'
According to the story, Aschenbrenner thought that the company's security measures against theft of its model weights and other important algorithmic secrets were woefully inadequate, and shared the security memo with several colleagues and his superiors. A few weeks later, a major security incident occurred. In the wake of the incident, Aschenbrenner shared the memo with several executives.
However, a few days later, it became clear that management was extremely unhappy that Aschenbrenner had shared the memo with the board, and that behind the scenes, the board had been pressuring management about security.
After that, the HR department formally warned Aschenbrenner about sharing the memo with the board of directors. Aschenbrenner said that the HR person said, 'Worrying about the Chinese Communist Party's espionage activities is racist and unconstructive.' After that incident, Aschenbrenner was fired at some point. The HR department clearly stated that the reason for his firing was 'because of the security memo.'
The official reason for his dismissal was said to be the sharing of security memos within the company, but OpenAI had told other employees that Aschenbrenner was suspected of being involved in leaking information, and the same information has been reported in
According to Aschenbrenner, OpenAI had a culture of sharing safety ideas with external researchers, and it was normal to get feedback from outside. Of course, important secrets were blacked out and internal links were modified to dead links, and Aschenbrenner often reviewed the information to make sure there were no confidential details.
One day, Aschenbrenner wrote a brainstorming document on future preparation, safety, and security measures for AGI (artificial general intelligence), shared it with three external researchers, and asked for their feedback. However, Aschenbrenner says that management pointed out that this act constituted information leakage. When Aschenbrenner pressed them to clarify what confidential information the document contained, management responded by saying that it contained information about the schedule of the company's internal plans, but Aschenbrenner says that the schedule was publicly available at the time.
Aschenbrenner also did not sign the letter calling for Sam Altman to return and for all directors to resign. The letter was signed by 90% of employees, but while Aschenbrenner thought it was appropriate to resign because the board had lost too much trust with employees, he took issue with a sentence in the letter that said 'we will not appoint independent outside directors.' Jan Reich and many other employees who did not sign later left the company. Aschenbrenner also speculates that Aschenbrenner may have incurred the resentment of management because, as part of his job, he had been urging management to keep their promises.
'The United States is a free country, and that's what I like about it. It makes sense for management to say, 'We're changing the company's policy and we don't agree with your views.' I joined OpenAI because I agreed with their philosophy, but that changed over time. That said, there are a lot of great people at OpenAI, and it's been a great honor to work with them,' he said.
Related Posts:
in Posted by log1p_kr