Tag: red_teaming

Media Monitor: OpenAI Develops AI Super-Hacker GPT-Red to Boost LLM Defenses

15 Jul 2026 1:09 p.m. · Source: MIT Technology Review

MIT Technology Review reports OpenAI has created GPT-Red, an AI super-hacker, to strengthen its language models like GPT-5.6 against cyberattacks through automated red-teaming and the discovery of new vulnerabilities.

Science & Technology Business Artificial Intelligence Computers and Internet OpenAI GPT-Red GPT-5.6 Cybersecurity Prompt Injection Red-teaming MIT Technology Review MIT science technology