New York City Council’s Committee on Technology heard sworn testimony from SpaceXAI on 16 October 2026 at a follow-up AI safety hearing, after the Council issued what it called its first subpoena on record against a private actor, according to a Council statement. Gabriela Castillo Madrid, the company’s Senior Manager of Global Government Affairs, appeared before councillors who had sought testimony from SpaceXAI at an earlier hearing.
Key points
- The Council subpoenaed SpaceXAI after seeking voluntary testimony, then threatened to seek judicial enforcement when the company missed the 5 October hearing.
- Castillo Madrid said a safety team could examine a sexualised image Speaker Julie Menin said she had made using Grok.
- A former employee alleged that three of SpaceXAI’s 200 technical employees worked on AI safety training during his time there.
- Councillors are considering a restriction on unauthorised AI depictions of elected officials and a financial incentive for whistleblowers.
SpaceXAI’s subpoena and the 5 October hearing
Speaker Julie Menin subpoenaed SpaceXAI on 28 September after the company did not agree to testify voluntarily, amNewYork reported. OpenAI, Anthropic, Google and Meta appeared under oath at the Council’s nearly 10-hour hearing on 5 October. SpaceXAI did not. Menin said during that hearing that the company was in “direct violation” of the subpoena and that the Council was pursuing enforcement in court.
The Council subsequently told SpaceXAI that it planned to ask a state Supreme Court judge to enforce the subpoena. SpaceXAI requested a meeting with Council officials on 7 October and later agreed to testify on 16 October, amNewYork reported. The threatened application gave the Council a means of seeking a judge’s order rather than relying on another invitation to the company.
AI Affairs covered the subpoena ahead of the 5 October AI safety hearing. At that hearing, the four companies that appeared gave few clear answers about catastrophic risks, according to amNewYork. None put a number on the likelihood of such an event, and they largely stopped short of saying they would automatically withhold a model that failed an independent safety review.
Menin questions Castillo Madrid about Grok
At the follow-up hearing, Menin told Castillo Madrid that she had created a sexualised image of herself within seconds using Grok. Menin asked how that was possible if the product had “strict controls”. Castillo Madrid could not speak to the images, the Council said, but told Menin her safety team “could look into that specific instance”. The answer offered to examine Menin’s example rather than account for the controls she had asked about.
The Council had also heard from British MP Jess Asato at the 5 October hearing. Asato is in a legal dispute with SpaceXAI after Grok created non-consensual explicit photos and videos using her likeness, according to the Council. Menin’s question on 16 October concerned an image she said she had produced herself, placing the product’s controls before the company representative in a different way.
Committee chair Carmen De La Rosa said the companies at the hearings had been unable to give councillors answers on privacy protections, harassment safeguards, whistleblower protections and risks from their technology operating outside intended controls. She said the committee had prepared follow-up letters and would continue pursuing legislation and advocacy. Her account was the Council’s assessment of the testimony, rather than a finding reached through enforcement proceedings.
Kim’s safety account and SpaceXAI’s response
Devin Kim, whom the Council described as a SpaceXAI whistleblower, alleged that he was terminated after raising concerns about Grok’s safety. He told councillors that, during his time at the company, only three of its 200 technical employees worked on AI safety training. That was Kim’s account of staffing during his employment, not a figure supplied by SpaceXAI for its workforce at the time of the hearing.
Asked about Kim’s termination and account, Castillo Madrid testified under oath that she was unaware of them, the Council said. Questions about the level of dedicated safety staff and the company’s view of Kim’s departure went unanswered at the hearing, MLex reported. MLex described the departure as a resignation. Kim alleged that he had been terminated.
Intro 504 and Intro-2605 before the Council
The Council is considering Intro 504, sponsored by Deputy Speaker Dr. Nantasha Williams. The proposal would prohibit unauthorised AI depictions of elected officials in photos, videos or audio. Williams said fabricated depictions could make officials appear to have said or done things they had not, putting public trust at risk when people seek information about their government.
The Council introduced Speaker Menin’s Intro-2605 ahead of the SpaceXAI hearing. The proposed measure would let whistleblowers receive a portion of fines or penalties recovered from AI companies that break the law, giving employees a financial incentive to report misconduct.
.jpg){kind=link}