GPT-Red: Unlocking Self-Improvement for Robustness

Summary

OpenAI describes GPT-Red, an automated red-teaming model trained with self-play reinforcement learning to find prompt injection vulnerabilities before deployment. The company says GPT-Red is used to adversarially train GPT-5.6, which it reports as its most robust model to prompt injections, with 6x fewer failures on its hardest direct prompt injection benchmark compared with its best production model from four months earlier.

Topics

• Prompt injection and jailbreaks

Related items

• LowFBI arrests Saudi OpenAI employee who allegedly planted recording device in conference roomSame vendor · CNBC Technology
• InfoAn OpenAI Employee Allegedly Used a Chatbot to Help Plant a Recording Device at the OfficeSame vendor · Wired (Security)
• InfoEmployee Arrested After Listening Device Found in OpenAI Conference Room, FBI and the Company SaySame vendor · SecurityWeek
• InfoOpenAI employee arrested after recording device found in officeSame vendor · The Guardian Technology
• InfoOpenAI engineer arrested for allegedly planting spy device in officeSame vendor · BleepingComputer