OpenAI 搞了个 GPT-Red 自动红队模型，提示注入赢人类 84% 对 13%，还发现了 Fake Chain-of-Thought 攻击。
OpenAI 训练了内部自动红队模型 GPT-Red，采用自我对弈强化学习对抗防御模型。在间接提示注入基准测试中，GPT-Red 以 84% 对 13% 的胜率击败人类红队。它还发现了一种新型攻击类别 "Fake Chain-of-Thought"。在 OpenAI 最难的直接注入基准上，GPT-Red 将 GPT-5.6 Sol 的失败次数减少了 6 倍。不过，OpenAI 承认该模型在多轮对话和基于图像的攻击方面仍有不足。
OpenAI Details GPT-Red: An Internal Automated Red-Teaming Model That Beat Human Red-Teamers 84% To 13% On Prompt Injection
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. The post 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 .