OpenAI Unveils GPT-Red AI Model That Automatically Finds Prompt Injection Vulnerabilities

OpenAI introduced GPT-Red, an automated red-teaming model designed to identify and exploit prompt injection vulnerabilities through self-play reinforcement learning. By generating adversarial prompts, this tool enhances the robustness of production models like GPT-5.6 Sol against unauthorized data access and system manipulation.