Tech

OpenAI President Greg Brockman Testifies on Volatile 2017 Meeting with Elon Musk and Internal Board Conflicts

Brockman recounts heated August 2017 encounter at Elon Musk's estate and subsequent efforts to remove board members amid concerns over absolute control

Author
Owen Mercer
Markets and Finance Editor
Published
Draft
Source: WIRED · original
‘I Actually Thought He Was Going to Hit Me,’ OpenAI’s Greg Brockman Says of Elon Musk
Former OpenAI president details fears of physical attack and leadership struggles over company control in federal trial

Greg Brockman, the former president of OpenAI, concluded his testimony in the Musk v. Altman federal trial on Tuesday by describing a volatile August 2017 meeting at Elon Musk's estate in Hillsborough, California. Brockman stated that during the encounter, Musk rejected a proposal for shared control of the organisation, leading Brockman to say he feared Musk would physically attack him. The testimony highlighted the intense friction between the company's leadership and Musk regarding the governance of the artificial intelligence venture.

The meeting took place at a 47-acre estate where Brockman and co-founder Ilya Sutskever had gathered to discuss the future of OpenAI. Prior to the discussion, Musk had gifted the pair new Tesla Model 3 cars, a gesture Brockman interpreted as an attempt to create a sense of indebtedness. During the negotiation, Musk insisted on absolute control over the company, which Brockman and Sutskever argued would amount to a dictatorship over the future of AI development.

Brockman recalled that after Musk rejected their offer for shared leadership, he stormed around the table and grabbed a painting Sutskever had gifted him. Musk threatened to cut off his funding to the nonprofit until Brockman and Sutskever resigned. Despite the hostility, Brockman testified that Musk later expressed interest in future collaborations, though the immediate atmosphere left the researchers feeling threatened by the Tesla CEO's erratic behaviour.

Beyond the personal confrontation, Brockman detailed internal deliberations where OpenAI leadership considered removing several board members due to Musk's desire for absolute control. He supported the removal or partial recusal of Adam D'Angelo, the CEO of Quora, following the launch of a competing chatbot in February 2023. Additionally, Brockman backed the removal of Helen Toner, who resigned in 2023 after assisting in the firing of Sam Altman.

The testimony also addressed the case of board member Shivon Zilis, who revealed she had given birth to Musk's twins via IVF, a fact Musk had not disclosed to her at the time. Although several board members wanted to remove Zilis, Brockman stated he convinced them to let her stay because she helped manage Musk's frustrations with the organisation. Zilis is expected to testify on Wednesday, though her appearance will not be livestreamed due to security concerns for her and her four children.

Brockman further illustrated Musk's unsuitability for leadership by recounting an incident where Musk insulted early AI chatbot technology to researcher Alec Radford. Musk told Radford the system was so stupid that a kid on the internet could do better, leaving the researcher demoralised. Brockman argued that this inability to see potential in early technology demonstrated why Musk was unfit to control an AI company.

The jury in the Musk v. Altman trial could begin deliberating on an advisory ruling as soon as next week. While OpenAI denies wrongdoing regarding Musk's donations, the testimony underscores the deep conflicts over governance and the influence Musk exerted during the company's formative years.

Continue reading

More from Tech

Read next: Microsoft disables dozens of GitHub repositories following open-source supply chain attack
Read next: Apple opens developer access to iOS, iPadOS and macOS 27 betas
Read next: Apple confirms macOS 27 Golden Gate requires Apple Silicon, ending Intel support