On October 12, 2026, WIRED published Reece Rogers’s account of tests in which basic prompts led Meta’s Muse AI agent to generate explicit sexual text.
What Meta said about adult conversations
Meta spokesperson Daniel Roberts told WIRED that Muse is for people 18 and older, that adult conversations are optional, and that the agent does not block legal adult behavior. He described those conversations as taking place in a private virtual machine accessible only to the user.
WIRED also reported that Meta pointed to Muse’s system prompt, which supports adult sexual expression and tells the agent to preserve the intensity selected by the user.
How Dots and Grok responded
In the same report, Rogers described similar prompts to OpenAI’s Dots and xAI’s Grok. Dots declined the prompts; Grok declined to sext and redirected the conversation toward tasks.
| Product | Response Rogers reported |
| Meta Muse | Generated explicit sexual text after basic prompts. |
| OpenAI Dots | Declined similar explicit prompts. |
| xAI Grok | Declined to sext and redirected the conversation toward tasks. |
Muse Image is a separate feature
Rogers’s test concerned Muse’s text chat. Separately, Rest of World described Muse Image as an image-generation feature and reported that Meta pulled it within 72 hours of launch.
The account-security concern
Calli Schroeder, senior counsel at the Electronic Privacy Information Center, warned that if an account holding explicit chats were hacked, the conversations could expose its user to embarrassment or extortion.
