2 ARTICLES TAGGED "GPT-RED"
As AI systems become more powerful, ensuring their safety is critical. Explore how GPT-Red automates red teaming to identify vulnerabilities like prompt injection and strengthen LLM security.
OpenAI has developed GPT-Red, an internal 'elite hacker' model designed to find and exploit vulnerabilities in other AI models. Currently locked for internal use, it represents a major leap in automated red-teaming and AI safety infrastructure.