Mumsnet Denies AI-Written Posts After 1 Prompt Leaks as Users Challenge OpenAI Explanation
Updated
Updated · The Guardian · Oct 7
Mumsnet Denies AI-Written Posts After 1 Prompt Leaks as Users Challenge OpenAI Explanation
1 articles · Updated · The Guardian · Oct 7
Summary
Justine Roberts said the leaked “AIBU Response Schema v.1.0” came from Mumsnet’s title-suggestion system after it sent a full reply prompt to OpenAI instead of a headline request.
The prompt surfaced under a post about a neurodivergent uncle and a sick child, instructing an AI to deliver a forum-style ruling and advice using Mumsnet shorthand such as AIBU, ND and DD.
Users called the incident “creepy” and “disappointing,” rejected Mumsnet’s shifting explanations, and argued the specificity of the prompt suggested an internal tool for drafting replies.
Roberts said Mumsnet has never tested AI to write posts, using it instead for coding, ad operations, conversational analysis and moderation flags, while stressing that “the people on it are real.”
The dispute lands as Mumsnet, founded in 2000, reported a 19% drop in monthly unique users to 6.1 million in 2025 and faces wider concern that AI-generated social content feels robotic and less trustworthy.