By JOHN F. McCARTHY/V.I. Free Press Staff
PALO ALTO, California – January 23, 2026 – A bipartisan group of 35 state and territory attorneys general, coordinated in part by U.S. Virgin Islands Attorney General Gordon C. Rhea, has sent a detailed letter to xAI calling for expanded measures to prevent its Grok AI chatbot from producing deepfake nonconsensual intimate images (NCII) and child sexual abuse material (CSAM).
Addressed to xAI’s safety team, the letter recognizes recent steps by the company—including technical restrictions on editing real people’s images into revealing clothing, content removals, investigations into potential violations, and reports to law enforcement—while expressing concern that these may not fully resolve the problem.
It cites media reports indicating persistent issues and urges xAI to go further as a market leader, given Grok’s direct integration with the X platform.The attorneys general highlight that Grok was designed with permissive features, including a marketed “spicy mode” for explicit content, which they argue facilitated abuse “at an astonishing scale.”
Users reportedly prompted the AI to “undress” women and children, generating sexualized alterations of real images—sometimes depicting minors in minimal clothing or explicit scenarios—and sharing them publicly on X.
Analyses cited in the letter suggest Grok produced thousands of such images per hour in late December/early January, often exceeding dedicated “nudify” tools.While noting that other AI platforms face similar challenges, the signatories emphasize Grok’s unique risks due to its social media tie-in and promotional framing of unrestricted generation as a feature.
They stress the severe harm to victims (overwhelmingly women and girls) and potential violations of state/federal laws on NCII and CSAM—even for synthetic or manipulated content.The letter requests prompt details on xAI’s plans to:
- Permanently block production of NCII (including suggestive but non-nude depictions) and CSAM.
- Remove existing harmful content.
- Suspend and, where appropriate, report offending users.
- Provide X users easy opt-out controls over Grok editing their images or responding to their posts.
- Ensure safeguards are robust and not merely paywalled.
The coalition expresses appreciation for prior discussions with xAI and a desire for ongoing dialogue, framing the request as an opportunity for leadership in responsible AI development.Signatories include attorneys general from Connecticut (William Tong), North Carolina (Jeff Jeff Jackson), Pennsylvania (Dave Sunday), Utah (Derek Brown), American Samoa (Gwen Tauiliili-Langkilde), Arizona (Kris Mayes), and others across 35 jurisdictions.This multi-state outreach follows heightened scrutiny of Grok, including California AG Rob Bonta’s separate investigation (announced January 14) and cease-and-desist letter (January 16) demanding an immediate halt to such content generation, citing potential criminal and civil violations under state law.xAI has previously stated via its
@Safety account that it prohibits illegal content, enforces policies against such prompts, and continues refining safeguards. No immediate public response to this specific letter has appeared.