XChallenges Brazil’s Order on Grok After Disputed Image-Generation Tests

Overview

• Brazil’s data protection, consumer and prosecutorial agencies gave X five days to implement and detail safeguards to stop Grok from generating sexualized images of children and adults without consent, warning of fines or legal action.
• In a filing, X said the technical note underpinning the measures omitted which Grok version, prompts and outputs were used, and it asked that the five-day clock begin only after full test procedures are disclosed.
• X asserted that the domain grokimagine.ai is not affiliated with its platform and requested suspension of the required measures if the tested images originated outside its official domains at grok.com or within X.
• Reuters verified that grokimagine.ai redirects to grokimaginex.ai, which features Grok-like branding and sends users to a third-party site, with Reuters unable to confirm any official integration with Grok.
• A Reuters test found one model blocked a bikini-edit prompt as a policy violation while another model labeled Smart produced the altered image, leaving the source of the problematic outputs unresolved.