Musk’s Leader for OpenAI Rival Was Recently Arrested for Domestic Violence
THE TAKEAWAY
• The arrest happened three days before Musk’s X.AI was incorporated and could now complicate the company’s efforts to recruit talent to challenge OpenAI.
Police in northern California last month arrested Igor Babuschkin, Elon Musk’s first known hire for a new company that aims to compete with artificial intelligence startup OpenAI, for domestic battery, public records show. A representative for the Santa Clara County District Attorney’s office, Lisa Buuck, said the office had reviewed the case and had no plans to charge Babuschkin with a crime.
The arrest, which took place in Palo Alto, Calif., in the heart of Silicon Valley, could complicate Babuschkin’s efforts to get the company off the ground and recruit talent. Babuschkin was an AI researcher at OpenAI as well as Alphabet’s DeepMind unit before joining Musk to create what Musk has referred to as a chatbot stripped of political correctness—an “anti-woke” version of OpenAI’s trailblazing ChatGPT, The Information reported in February.
Babuschkin and Musk didn’t respond to requests for comment.
Around 9:30 a.m. on March 6, local police responded to a report of a domestic violence battery at 588 Webster St., a luxury apartment building in downtown Palo Alto, near Stanford University, according to public records and a police spokesman. The alleged victim has not been publicly identified.
“Following an investigation, police arrested the suspect (Igor Babuschkin) for misdemeanor domestic violence and booked him into the Santa Clara County Main Jail,” Palo Alto Police Department spokesperson Brian Philip said in an email. Bail was set at $10,000 and he posted bond and was released at 5:46 p.m. that day, Philip said.
Three days later, X.AI, Musk’s new company, was incorporated in Nevada, according to state filings. X.AI is part of the same Musk holding company that owns Twitter.
Babuschkin and Musk have discussed assembling a team to pursue AI research to improve the “reasoning abilities and the factualness of these language models,” Babuschkin previously told The Information, referring to machine-learning models that can generate elaborate text in response to simple prompts. Musk told Fox News host Tucker Carlson in a recent interview that he was “going to start something, which I call ‘Truth GPT,’ or a maximum truth-seeking AI that tries to understand the nature of the universe.” (GPT stands for generative pretrained transformer, a type of machine-learning model known as a large-language model that is behind the OpenAI chatbot.)
Musk’s Criticisms
Musk co-founded and previously funded OpenAI as a counterweight to Google and its Alphabet subsidiary, DeepMind. Musk hinted in a recent tweet that he may sue OpenAI because it started as a nonprofit and later created a for-profit unit that has raised billions of dollars from Microsoft—a development Musk recently said went against his original intentions. In 2018, Musk parted ways with OpenAI and tried to recruit some OpenAI researchers to join him at another company he runs, Tesla. (Ironically, Musk told employees at OpenAI, which was then still a nonprofit, that Tesla had a better shot at developing complex software because it was a for-profit firm with access to capital.)
Musk has since criticized ChatGPT for producing results that appeared to avoid inflammatory subjects or that reflected a left-leaning political bias. OpenAI President Greg Brockman recently told The Information that such criticisms were “legitimate” and “the system we implemented did not reflect the values we intended to be in there.” He said OpenAI staff “want the default personality of OpenAI to be one that treats all sides equally.”
At DeepMind, Babuschkin was co-author of a 2019 paper demonstrating a machine-learning model that could play the videogame StarCraft II at the same level as the game’s highest-ranked humans. He left DeepMind in February. Babuschkin also previously worked at OpenAI for a year and a half between 2020 and 2022.
Researchers at DeepMind and OpenAI have created software that can ace videogames as a step toward developing software that can solve more practical tasks for humans. Such advances have paved the way for apps, including ChatGPT, that can solve more-practical tasks for humans. The companies’ ultimate goal is to create artificial general intelligence, software systems that can think and learn on their own. For now, DeepMind is putting aside longstanding grudges with Google as the two Alphabet units jointly develop software that could rival ChatGPT.