kapynPolicy & Regulation

Hundreds asked ChatGPT for poison and bioweapon recipes and some got step-by-step high school level guides

GPT-5's internal risk rating sparked debate after reports revealed it provided dangerous instructions to users. OpenAI flagged the model as high-risk in summer 2025 before downgrading its threat level that fall. The incident highlights ongoing challenges in aligning frontier models and managing safety thresholds during deployment.

The Decoder·Jul 26, 2026

Opening Kapyn…