SpaceX artificial intelligence policy leader Gabriela Castillo Madrid testified before the New York City Council on Friday, defending the company's safety measures for its Grok model while facing pointed criticism over the generation of non-consensual sexualized images. The hearing, conducted via video, marked SpaceX's first appearance before the council after initially evading a subpoena and threatening legal enforcement.
Council Speaker Julie Menin challenged Castillo Madrid's assertion that appropriate safeguards were in place. Menin stated that she and her team had discovered a sexualized image of herself on Grok within seconds of searching for her name. "I'm going to completely push back on that," Menin said. "It's just unacceptable." In response, Castillo Madrid stated that the company maintains strict controls and would investigate the specific instance.
Carmen De La Rosa, who leads the council's tech committee, described the ability to easily sexualize images of real women without consent as disgusting. She questioned how the company could claim to operate safe models given these capabilities. The council opened its investigation following intense public debate regarding the risks of advanced AI models, with the aim of using testimony to guide new regulations.
Castillo Madrid, who joined SpaceX in government affairs earlier this year, previously worked at X, which is now owned by SpaceX, and spent nearly three years at the U.S. Department of Commerce. When asked about employee departures, she said she was not aware of any fired whistleblowers. Menin countered by citing Devin Kim, a former xAI employee reportedly fired after raising safety alarms. Castillo Madrid stated she was unaware of that specific case but described a company culture where employees can voice concerns.
Devin Kim, now President of the nonprofit Center for AI Safety, also testified via video. Kim stated he built systems used to train and evaluate Grok and began raising concerns in January 2025, which he said were dismissed by his manager. Kim claimed that under prior standards at the company, tasteful nudity was considered acceptable. He noted that only three employees worked on AI safety during his tenure, many of whom left after his resignation due to frustration with how safety was handled.
The council also addressed other potential misuses of Grok. Councilmember Mercedes Narcisse questioned why the company promoted uploading X-rays to the model and whether HIPAA protections applied. Castillo Madrid responded that Grok is not meant to be a clinician and that users decide if their conversations are used for model training. Narcisse criticized the vagueness of these replies and warned against expecting definite health answers from the platform.
The hearing followed a February merger in which Elon Musk combined SpaceX with xAI, giving the aerospace and defense tech firm control of Grok. The model is used by some federal agencies, including the Pentagon. SpaceX went public in June with a valuation of approximately $2 trillion and completed a $60 billion acquisition of AI coding startup Cursor in August.
The initial avoidance of the council hearing also drew criticism from U.K. Parliament member Jess Asato, who is suing xAI over the creation and distribution of sexualized images of her without consent. Asato alleged breaches of European data protection rules and criticized SpaceX for its refusal to appear last week. SpaceX did not respond to a request for comment ahead of the meeting.