OpenAI Hardens ChatGPT Atlas Against Prompt Injection
OpenAI is enhancing ChatGPT Atlas's defenses against prompt injection attacks using reinforcement learning-based automated red teaming. This proactive discover-and-patch cycle aims to identify novel vulnerabilities as AI becomes more agentic.
