Elon Musk's SpaceXAI finally faced New York City lawmakers on Friday, Oct. 16, and the hearing turned tense almost at once. City Council Speaker Julie Menin said she had typed her own name into Grok, the company's AI chatbot, that morning and found a sexualized image of herself "within a second."
The company only showed up after the Council issued a subpoena and threatened to have a judge enforce it.
Key takeaways
- SpaceXAI testified under oath before the Council's Committee on Technology after skipping an Oct. 5 hearing.
- Menin said Grok surfaced a sexualized image of her within seconds, despite the company saying Grok has "strict controls."
- The company's witness couldn't say how many people lead safety, what insurance covers catastrophic AI harm, or whether any whistleblowers had been fired.
- The Council is weighing bills that would require independent safety checks before AI can be deployed in the city.
"How Is It That Within a Second…?"
Testifying by video was Gabriela Castillo Madrid, SpaceXAI's senior manager of global government affairs. She told the committee that "safety is core to how we build Grok" and that the chatbot has "strict controls" against sexualized images of real people, CNBC reported.
Menin pushed back, holding up her phone. "I'm going to completely push back on that," she said. "Just this morning, myself and my team, we put our own names, my own name, into Grok, and there's absolutely a sexualized image of myself."
Castillo Madrid said she couldn't speak to that specific case but that the company's safety team "could look into" it, according to the Council's account of the hearing.
Committee chair Carmen De La Rosa called it "disgusting" that Grok lets users sexualize images of real women without their consent.
Questions She Couldn't Answer
Lawmakers left with a list of follow-ups. According to amNY and Fox News, Castillo Madrid:
- said "everybody is in charge of safety" when asked how many people lead safety work, then agreed to send the Council a number later
- said she didn't know the details of the company's insurance for catastrophic AI events, though the company takes "catastrophic risk very seriously"
- told De La Rosa she wasn't aware of any whistleblowers being fired
- said Grok is "not meant to be a clinician, a doctor," after Council Member Mercedes Narcisse criticized users being encouraged to upload X-rays and MRIs for a diagnosis
"You're supposed to give us answers," Narcisse told her.
The hearing also heard from Devin Kim, a former xAI employee who says he was fired after raising safety concerns about Grok. He appeared by video with his lawyer. Castillo Madrid testified under oath that she didn't know about his firing.
- 1stCouncil subpoena to a private companyIssued to SpaceXAI on Sept. 28
- ~10 hrsLength of the Oct. 5 hearing
- 46Council members who took part on Oct. 5
- 4AI firms that testified on Oct. 5Anthropic, OpenAI, Google and Meta
Source: New York City Council
How It Took a Subpoena
The Council's AI safety probe started in September. Menin sent letters to five AI companies between Sept. 15 and 17 asking them to testify. Four agreed. SpaceXAI, the company formed in February when Musk merged SpaceX with xAI, didn't respond.
- Sept. 15–17Letters go outMenin asks five AI companies to testify about safety.
- Sept. 28SpaceXAI subpoenaedThe Council's first subpoena to a private company, according to the Council.
- Oct. 5Marathon hearingAnthropic, OpenAI, Google and Meta testify under oath. SpaceXAI doesn't show up.
- Oct. 7Company asks to meetAfter the Council says it will ask the New York State Supreme Court to enforce the subpoena.
- Oct. 16SpaceXAI testifiesCastillo Madrid appears by video before the Committee on Technology.
"It took a subpoena and the threat of continued legal action to compel SpaceXAI to show up and testify before New Yorkers today, and yet their testimony was simply unacceptable," Menin said after the hearing.
What the Council Wants to Do About AI
The Council published a slate of AI bills ahead of the Oct. 5 hearing, The Next Web reported. They include:
- a Menin bill barring companies from selling or deploying an AI system in the city without third-party validation
- a Menin bill giving whistleblowers a share of fines recovered from AI companies
- a requirement that city contractors report AI safety incidents within 24 hours
Lawmakers have also discussed letting New Yorkers harmed by AI take legal action.
FAQ
Why did the NYC Council subpoena SpaceXAI?
SpaceXAI was the only one of five AI companies that didn't respond to the Council's request to testify at its Oct. 5 AI safety hearing. Speaker Julie Menin issued a subpoena on Sept. 28 to compel testimony under oath.
What did Julie Menin say about Grok?
Menin said she and her team typed her name into Grok on the morning of Oct. 16 and found a sexualized image of her "within a second," even though SpaceXAI said Grok has strict controls against such images of real people.
Who testified for SpaceXAI?
Gabriela Castillo Madrid, SpaceXAI's senior manager of global government affairs, testified by video. She joined SpaceX's government affairs team this year after working at X and the U.S. Commerce Department, according to CNBC.
What happens next?
Castillo Madrid promised written follow-up answers. The Council says it will use testimony from both hearings to shape AI safety legislation.