OpenAI safety team flagged teen planning school shooting, lawsuits allege inadequate response
Sarah, a former member of OpenAI’s intelligence and investigations team, spent weeks reviewing flagged ChatGPT conversations. She discovered a teenager who repeatedly discussed bullying, plotted a school shooting, and set a date that passed without incident. The logs sat in a backlog that could stretch months, leaving her unable to intervene in time.
Key points
- Sarah, an OpenAI intelligence reviewer, saw a teenager plan a mass shooting in ChatGPT logs that sat in a months‑old backlog.
- Eight US states sued OpenAI in October, accusing it of misleading safety claims and failing to act on violent user conversations.
- The Tumbler Ridge shooter’s flagged chats were not reported to police; OpenAI later apologized and said a revised protocol will refer such cases.
The investigation notes that eight U.S. states sued OpenAI in October, accusing the company of misleading the public about ChatGPT’s safety and of failing to act on violent user behavior. Additional lawsuits—over 30 by families alleging deaths linked to the bot and 45 by survivors of school shootings in Florida and Canada—claim OpenAI did not alert police despite internal alerts. A separate case in Tumbler Ridge, Canada, involved a shooter whose flagged chats were not reported, prompting an apology from CEO Sam Altman and a promise of revised protocols.
OpenAI’s safety team, described as “small” with roughly a dozen to low‑dozens of staff, contrasts with the billions of weekly ChatGPT users and the much larger moderation forces at Meta, TikTok and Discord. The article highlights ongoing concerns that conversational AI’s design—flattering, memory‑rich, and emotionally supportive—may reinforce harmful beliefs and pose immediate public‑safety risks.
The story so far
4 episodes →- OpenAI safety team flagged teen planning school shooting, lawsuits allege inadequate responsethis story
‘If it wasn’t for ChatGPT my brother would still be alive’: inside OpenAI’s safety unit
observer.co.uk · 1 November 2026
Loading the full article…
This text was published by observer.co.uk and written by Patricia Clarke, Katie Gunning. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Reddit discussionreddit.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Policy & Regulation
All →- Anthropic restricts Claude Code from recreating specific video games · 2 src
- Anthropic updates Claude to block AI-generated game clones and reverse engineering · 8 src
- OpenAI, Google, Microsoft, and Anthropic back shared robots.txt opt-out for AI training · 11 src
- Amazon Japan adds agent policy to seller terms · 1 src
- Microsoft warns North Korean hackers use AI to scale phishing and malware · 2 src
Comments
via GitHub Discussions