Story perspectives

xAI Sues South Carolina Man Over Grok Deepfakes

7/16/2026

Summary

xAI Sues Harwood

xAI sued Terry Harwood in Texas on Tuesday, a South Carolina resident who faces eight felony child-sexual-abuse charges.
The complaint says Harwood uploaded non-sexual images of adults and minors to Grok and used the chatbot to create explicit deepfake content.
xAI seeks damages and a ban on Harwood’s use of Grok, saying it suspended 52,222 accounts, filed 73,604 NCMEC reports in 2026, and those led to at least 244 arrests.

Summary

Harwood Lawsuit

The Harwood lawsuit joins a wave of AI sexual-abuse cases, including a March Tennessee teen class action and Ashley St. Clair’s claim that Grok turned childhood photos into sexualised images.
Baltimore sued xAI for alleged consumer-protection violations tied to its AI service.
Australia’s eSafety Commissioner Julie Inman Grant said complaints about Grok’s “edit image” feature doubled since January, and fined a Gold Coast man for AI-generated pornography, a case courts may watch.

Summary

xAI Under Investigation

Regulators in California, the U.K., the European Union (EU) and Ireland opened probes into xAI’s Grok after early-January 2026 reports of sexualized deep-fake generation.
Harwood bypassed new safeguards and was arrested on March 9, 2026 on three second-degree and five third-degree minor sexual exploitation counts.
xAI’s lawsuit seeks monetary damages and reimbursement for legal costs, noting Harwood repeatedly altered prompts after Grok refused.

Summary

Researchers Bypass Safeguards

Researchers discovered a jailbreak prompt that bypasses Grok’s image safeguards, enabling sexual or violent outputs without direct requests.
Security experts say the prompt’s indirect language evades safety filters, exposing guardrail weaknesses and appearing as fast as moderation updates, creating a cat-and-mouse battle.
Google, Anthropic and other leaders announced stricter image-generation policies, while analysts warn technical fixes may be insufficient and urge regulators to consider mandatory legal actions against abusers.

History

Historical Context

The 2018 Senate hearing led by Senator Mark Warner warned that over 1,000 AI-generated porn videos were circulating, prompting early bans on deepfakes and the Malicious Deep Fake Prohibition Act. By early 2026, analyses of xAI’s Grok showed 2% of 20,000 images depicted minors and users produced 6,700 sexualized images per hour, fueling the current lawsuit and global regulatory push.

1 of 5

xAI Sues Harwood

xAI sued Terry Harwood in Texas on Tuesday, a South Carolina resident who faces eight felony child-sexual-abuse charges.
The complaint says Harwood uploaded non-sexual images of adults and minors to Grok and used the chatbot to create explicit deepfake content.
xAI seeks damages and a ban on Harwood’s use of Grok, saying it suspended 52,222 accounts, filed 73,604 NCMEC reports in 2026, and those led to at least 244 arrests.

1 / 5