OpenAI News·· 2026-03-10AI 评分37
OpenAI IH-Challenge 提升前沿 LLM 指令层次结构
Improving instruction hierarchy in frontier LLMs
AI 导读
OpenAI 推出的 IH-Challenge 训练模型优先执行可信指令，提升指令层次结构，增强安全可控性和抵抗提示注入攻击的能力。
原文
IH-Challenge trains models to prioritize trusted instructions, improving instruction hierarchy, safety steerability, and resistance to prompt injection attacks.
来源：OpenAI News · openai.com