OpenAI ships GPT-Red to train GPT-5.6 against prompt injection attacks
The internal-only model uses adversarial self-play to generate attacks that harden OpenAI's production models before deployment.
By Ryan Merket · Published
Primary source: X / OpenAI
Why it matters
Prompt injection is becoming an infrastructure problem for agent builders. OpenAI's GPT-Red turns red teaming into a training loop, but the public evidence still rests on OpenAI's own benchmarks.
OpenAI introduced GPT-Red on July 15th, an internal automated red-teaming model built to find prompt injection vulnerabilities at scale and feed those attacks back into the training of production models.…