Researchers Propose a Better Way to Report Dangerous AI Flaws
AI researchers discovered a glitch in GPT-3.5 that led to incoherent output and exposure of personal information.
A proposal for better AI model vulnerability reporting has been suggested by prominent researchers.