
GPT-Red: OpenAI's internal 'super-hacker' AI red-teamer and what it means for AI safety and builder teams
Published by AINave Editorial • Reviewed by Ramit
OpenAI detailed GPT-Red, an internal AI system that uses self-play reinforcement learning to attack its own models and surface prompt injection vulnerabilities before they reach users. The system achieved an 84% success rate across scenarios compared to 13% for human red-teamers, and its findings have been used to harden GPT-5.6 against a class of attacks that previously succeeded over 95% of the time.
What happened
GPT-Red is an automated red-teaming model trained via self-play reinforcement learning. It acts as an attacker against defender models across varied scenarios, earning rewards for successful exploits while defenders are rewarded for holding firm. As defenses improve, the attacker invents harder attacks and the loop repeats. The company said the approach outpaces human red-teamers: GPT-Red succeeds on 84% of scenarios versus 13% for humans, and it cuts direct prompt injection failures to a sixth of the rate in its best production model from four months earlier.
A class of "fake chain-of-thought" attacks that worked more than 95% of the time against GPT-5.1 now succeeds less than 10% of the time against GPT-5.6. The tool has also broken autonomous agents, including [hijacking a Vendy
Sources
- OpenAI details GPT-Red, an AI that attacks its own models to find flaws - SiliconANGLE
- GPT-Red: Unlocking Self-Improvement for Robustness | OpenAI
- OpenAI Built an AI to Attack Itself: GPT-Red Exposed Flaws Humans...
- Thread By @OpenAI - Introducing GPT-Red An internal...
- OpenAI's GPT-Red Explained: Automated Red-Teaming Where AI...
- Meet GPT-Red, OpenAI's new AI hacker that finds flaws in its models before release
- OpenAI built an AI to attack itself: GPT-Red exposed flaws humans missed
- OpenAI Uses AI Red Team to Strengthen GPT-5.6 Against Prompt Injection Attacks
- OpenAI built an AI super-hacker to break its own models, then locked it away
- OpenAI details GPT-Red, an internal automated red-teaming model...
- OpenAI Rolls Out Powerful New GPT-5.6 Models—But Limits Users After Government Request
- After Mythos 5 ban, OpenAI brings GPT-5.5-Cyber to find software bugs humans miss
- OpenAI expands Daybreak with Patch the Planet and full GPT-5.5-Cyber release
- OpenAI Has New AI Models. Here’s Why You Can’t Use Them
- OpenAI and Anthropic limit new AI models to Trump-approved customers during cybersecurity review






















