SpaceXSI Faces City Council Over Grok Chatbot Failures
A hearing on AI safety revealed how Grok generates prohibited sexualized images despite corporate claims of strict controls.
Updated on Oct. 16, 2026 in Artificial Intelligence
Live Poll
Do you trust AI companies to effectively self-regulate the safety of their image-generation tools?
During a New York City Council hearing on October 16, 2026, a SpaceXSI representative testified on the safety protocols of the company's Grok chatbot. The session included a live demonstration where Council Speaker Julie Menin generated a sexualized image of herself using the platform, contradicting claims of strict output controls.
Why it matters
This hearing highlights the tension between AI developers and municipal oversight regarding the proliferation of non-consensual sexualized imagery. It underscores the challenges regulators face in enforcing safety standards against platform-level moderation failures.
The Center for Countering Digital Harm estimated that Grok produced 3 million photorealistic sexualized images within 11 days. SpaceXSI, which currently maintains a spicy mode for queries rejected by other systems, asserts that these outputs stem from bad actors bypassing existing controls.
The players
Gabriela Castillo Madrid
The SpaceXSI representative who testified before the New York City Council.
Julie Menin
The New York City Council Speaker who demonstrated the chatbot's generation of sexualized imagery.
SpaceXSI
A technology company currently litigating AI regulations in California, Colorado, and Minnesota.
Center for Countering Digital Harm
The research organization that provided the data on Grok image generation volume.
The details
Grok users access image generation via a paid X subscription, which features a 'spicy mode' designed to process queries typically rejected by other AI models. SpaceXSI claims these systems include strict controls against generating sexualized imagery of real people, though testimony suggests these guardrails can be bypassed. The company also maintains internal safety initiatives, reporting over 100,000 submissions to the National Center for Missing & Exploited Children, which it claims have led to nearly 300 arrests.
Timeline
December 29, 2025: SpaceXSI launched the Grok image-generation feature.
October 9, 2026: Other technology firms testified before the City Council.
October 16, 2026: Gabriela Castillo Madrid testified before the Committee on Technology.
The Tech Race
This hearing sits within the broader conflict between SpaceXSI and state-level AI regulations, where the company is currently challenging mandates in California, Colorado, and Minnesota. The discrepancy between official safety claims and the generation of prohibited imagery mirrors a wider struggle over industry-wide moderation benchmarks.
The Grok image-generation feature remains accessible to any user with a paid X subscription. Residents should note that while the company reports these outputs result from circumvention by bad actors, the platform's current safety filters have not prevented the widespread creation of non-consensual sexualized content.
The takeaway
The tension between SpaceXSI and municipal regulators suggests that current moderation guardrails are insufficient to prevent the generation of harmful imagery. Observers should track upcoming council deliberations for potential new local mandates or subpoena outcomes that may force stricter compliance from AI developers.
Further reading
For broader context on current industry standards and regulatory challenges, see Artificial Intelligence.
Source note: This article includes information reported by La Voce di New York.
Live Poll
Do you trust AI companies to effectively self-regulate the safety of their image-generation tools?
