In a newly launched deposition filed in Elon Musk’s case in opposition to OpenAI, the tech govt attacked OpenAI’s security report, claiming that his firm, xAI, higher prioritizes security. He went as far as to say that “No one has dedicated suicide due to Grok, however apparently they’ve due to ChatGPT.”
The remark got here up in a line of questioning a few public letter Musk signed in March 2023. In it, he referred to as on AI labs to pause improvement of AI methods extra highly effective than GPT-4, OpenAI’s flagship mannequin on the time, for not less than six months. The letter, which was signed by over 1,100 individuals, together with many AI consultants, acknowledged there was not sufficient planning and administration going down at AI labs, as they had been locked in an “out-of-control race to develop and deploy ever extra highly effective digital minds that nobody — not even their creators — can perceive, predict, or reliably management.”
These fears have since gained credibility. OpenAI now faces a sequence of lawsuits alleging that ChatGPT’s manipulative dialog techniques have led a number of individuals to expertise destructive psychological well being results, with some dying by suicide. Musk’s remark means that these incidents may very well be used as fodder in his case in opposition to OpenAI.
The transcript of Musk’s video testimony, which came about again in September, was filed publicly this week, forward of the anticipated jury trial subsequent month.
The lawsuit in opposition to OpenAI facilities on the corporate’s shift from a nonprofit AI analysis lab to a for-profit firm, which Musk claims violated its founding agreements. As a part of his arguments, Musk claims that AI security may very well be compromised by OpenAI’s industrial relationships, as such relationships would place pace, scale, and income above security considerations.
Nonetheless, since that recording, xAI has confronted security considerations of its personal. Final month, Musk’s social community X was flooded with nonconsensual nude pictures generated by xAI’s Grok, a few of which had been stated to be of minors. This led the California Lawyer Basic’s workplace to open an investigation into the matter. The EU can also be operating its personal investigation, and different governments have taken motion, too, with some imposing blocks and bans.
Within the newly filed deposition, Musk claimed he had signed the AI security letter as a result of “it appeared like a good suggestion,” not as a result of he had simply included an AI firm trying to compete with OpenAI.
“I signed it, as many individuals did, to induce warning with AI improvement,” Musk stated. “I simply needed … AI security to be prioritized.”
Musk additionally responded to different questions within the deposition, together with these about synthetic basic intelligence, or AGI — the idea of AI that may match or surpass human reasoning throughout a broad vary of duties — saying “it has a threat.” He additionally confirmed that he “was mistaken” about his supposed $100 million donation to OpenAI; the second amended criticism within the case places the precise determine nearer to $44.8 million.
He additionally recalled why OpenAI was based, which, from his perspective, was as a result of he was “more and more involved concerning the hazard of Google being a monopoly in AI,” including that his conversations with Google co-founder Larry Web page had been “alarming, in that he didn’t appear to be taking AI security critically.” OpenAI was fashioned as a counterweight to that menace, Musk claimed.
