In a lawsuit against OpenAI, Elon Musk slammed OpenAI's security record in the latest public trial testimony, claiming that his xAI takes security more seriously, and said harshly: "No one committed suicide because of Grok, but obviously someone committed suicide because of ChatGPT."
The statement came as lawyers questioned Musk about an open letter. This open letter was released in March 2023. Musk was one of the signatories at the time, calling on major AI laboratories to suspend the development of systems more powerful than OpenAI's flagship model GPT-4 for at least six months. The open letter, signed by more than 1,100 people, including multiple AI experts, believes that major laboratories, without adequate planning and management, have fallen into an "out of control race" to develop and deploy increasingly powerful "digital minds", and these systems "cannot even be understood, predicted or reliably controlled by their creators."
In subsequent developments, these concerns gained more realistic footing. OpenAI is currently facing multiple lawsuits, alleging that ChatGPT contains manipulative words in conversations, which has led to serious mental health problems for many users, some of which ended in tragedy of suicide. Musk directly linked the "suicide incident" to ChatGPT, which was interpreted as an attempt to use this as a bargaining chip to attack OpenAI in the lawsuit.
Musk's video testimony was recorded in September last year, and the transcript was submitted as a public document this week, paving the way for a jury trial scheduled to begin next month. The core of this lawsuit points to OpenAI's transformation from a non-profit AI research institution to a profit-oriented corporate structure. Musk claimed that this change violated the relevant agreements when the company was founded. He further argued that cooperation with large commercial partners and commercialization paths would allow OpenAI to prioritize speed, scale and revenue over safety in decision-making, thereby compromising the priority of AI safety.
However, since the testimony was recorded, Musk’s own xAI and its product Grok have also been embroiled in safety controversies. In January this year, a large number of non-consensual nude photos generated by xAI’s Grok appeared on Musk’s social platform The incident has attracted great attention from regulatory authorities, and the California Attorney General’s Office has launched an investigation into xAI and Grok. The European Union has also launched an independent privacy investigation, and some national and regional regulatory agencies have taken measures such as blocking or restricting access to deal with this torrent of "deep fake" nude photos.
In the latest public testimony, Musk said that he signed the AI security open letter that year because it "seemed like a good idea," not because he had just registered an AI company that was preparing to compete with OpenAI. "I signed it, as many people did, to urge caution in the development of AI." Musk said, "I just hope that... AI safety can be prioritized."
Musk was also asked a number of questions related to the prospects of AI during his testimony, including his views on "artificial general intelligence" (AGI). AGI generally refers to AI systems that can match or exceed human reasoning capabilities on a wide range of tasks. Musk said in his testimony that AGI "presents risks." He also admitted that his previous repeated mention of donating "$100 million" to OpenAI was inaccurate. The second amended indictment in the case, based on relevant financial records, estimated Musk’s actual donation to OpenAI to be approximately US$44.8 million.
When talking about the original intention of establishing OpenAI, Musk recalled that he promoted the creation of this organization in part because of "the growing concern about the danger of Google forming a monopoly in the field of AI." He said that the many conversations he had with Google co-founder Larry Page about AI issues were "shocking" because Page "didn't seem to take AI security issues seriously." In Musk's narrative, OpenAI is envisioned as a "hedging force" to check and balance this potential threat.