SpaceX accepts the NYC Council's AI subpoena after a threat of court enforcement

Whether the threat worked is, for now, the council's claim (or rather the council's sequence of events, since SpaceX has not described its own). SpaceX did not respond to a request for comment ^[8], so the only account of why it changed course is Council Speaker Julie Menin's: she said SpaceX agreed to appear two days after the council told it that it intended to seek judicial enforcement ^[2]. The company had seven days between receiving the subpoena and the hearing it skipped ^[16], and, according to amNewYork, asked for a meeting on Oct. 7 ^[7], two days after that hearing ^[15]. Neither source dates the council's warning itself.

We think the threat worked because refusing was expensive for SpaceX in particular. It went public in June and is valued at over $2 trillion ^[9], and its Grok models are used by some federal agencies, including the Pentagon ^[10]. Carmen De La Rosa, who chairs the tech committee, made that federal footprint her argument. "With the proliferation of Elon Musk's technology at the federal level, it is even more important they come to the discussion with robust context about data privacy, safeguards, and whistleblower protections," she said ^[11]. Neither side has gone to court, and there is no report that SpaceX contested the subpoena. The counter-thesis is that the agreement shows only how SpaceX weighed the cost of a fight, and leaves the council's authority as untested as it was.

Attendance is the lower bar. Four companies, Anthropic, OpenAI, Google and Meta ^[17], cleared it on Oct. 5 ^[12], and SpaceX's testimony on Friday will be read against theirs. Menin said in her statement that New Yorkers "deserve to hear directly from the companies developing this powerful technology about the safeguards they are putting in place to protect people from harm and how they will be held accountable when those safeguards fail" ^[13].

SpaceX's testimony can give the answers the other four gave, in which case the threat produced an appearance and little else. It can offer a commitment the others declined, such as pulling a model that fails an independent safety review, in which case the threat achieved more than we credit it with. Or the arrangement can fray before Friday and the council can go to a judge after all. A filing would be the first test of whether a state judge enforces a city subpoena against an AI company. We expect the first outcome. A commitment from SpaceX that the other four refused would prove us wrong.