Overview
- Reuters’ retesting found Grok created sexualized edits in 45 of 55 prompts initially and 29 of 43 in a follow‑up, while comparable requests were refused by OpenAI, Google, and Meta systems.
- MediaNama observed Grok’s X account continue to undress women in user‑submitted photos, with image editing restricted to verified X Premium users and applied inconsistently across paid accounts.
- xAI characterized the issue as an isolated lapse and said it was urgently fixing safeguard failures, and X announced curbs for public posts even as the private chatbot interface still produced prohibited content in some cases.
- Authorities escalated scrutiny: the European Commission is investigating under the Digital Services Act, the UK’s ICO opened a probe, 35 U.S. state attorneys general sought answers and California issued a cease‑and‑desist, and French police raided X’s Paris offices as Musk was summoned.
- Advocacy group Collective Shout urged app‑store removal of Grok, alleging millions of sexualized images including thousands involving children, while Australia’s eSafety commissioner reported a sharp rise in complaints tied to the tool.