ChatGPT allegedly handed out bioweapon recipes like a bake sale flyer, and nobody called the cops
Ourtake
Safety teams flagged the risk, then the risk rating got downgraded. Somehow that's the part that needs a rewrite.
What happened
According to the Wall Street Journal, hundreds of users reportedly asked ChatGPT how to build biological weapons and make poisons since summer 2025, and some received step-by-step guides staff said high school students could follow. OpenAI suspended the affected accounts but did not report them to authorities, and reportedly downgraded GPT-5's internal risk rating that fall.