{"slug": "openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human", "title": "OpenAI Details GPT-Red: An Internal Automated Red-Teaming Model That Beat Human Red-Teamers 84% To 13% On Prompt Injection", "summary": "OpenAI developed GPT-Red, an automated red-teaming model that outperformed human red-teamers 84% to 13% on prompt injection attacks, discovering a novel 'Fake Chain-of-Thought' attack class and reducing GPT-5.6 Sol's failures sixfold on a direct injection benchmark. The model, trained via self-play reinforcement learning, still struggles with multi-turn and image-based attacks.", "body_md": "OpenAI trained GPT-Red, an internal-only attacker model, using self-play reinforcement learning against a population of defender LLMs. It beat human red-teamers 84% to 13% on a replicated indirect prompt injection arena, found a novel \"Fake Chain-of-Thought\" attack class, and cut GPT-5.6 Sol's failures 6x on OpenAI's hardest direct injection benchmark. OpenAI concedes it still struggles with multi-turn and image-based attacks.\n\nThe post [OpenAI Details GPT-Red: An Internal Automated Red-Teaming Model That Beat Human Red-Teamers 84% To 13% On Prompt Injection](https://www.marktechpost.com/2026/07/16/openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human-red-teamers-84-to-13-on-prompt-injection/) appeared first on [MarkTechPost](https://www.marktechpost.com).", "url": "https://wpnews.pro/news/openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human", "canonical_source": "https://www.marktechpost.com/2026/07/16/openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human-red-teamers-84-to-13-on-prompt-injection/", "published_at": "2026-07-16 18:48:04+00:00", "updated_at": "2026-07-16 19:10:19.417780+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-safety", "ai-research", "large-language-models", "ai-agents"], "entities": ["OpenAI", "GPT-Red", "GPT-5.6 Sol", "MarkTechPost"], "alternates": {"html": "https://wpnews.pro/news/openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human", "markdown": "https://wpnews.pro/news/openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human.md", "text": "https://wpnews.pro/news/openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human.txt", "jsonld": "https://wpnews.pro/news/openai-details-gpt-red-an-internal-automated-red-teaming-model-that-beat-human.jsonld"}}