首页 / 大模型 LLMs respond differently to harmful prompts when AI watermarking is used 大模型 2026-09-18 来源：Ars Technica 自动聚合 阅读 1 次 · 访客 1 位 SynthID can cause models to follow harmful instructions they would otherwise refuse.] # Ars Technica# llm 阅读原文（Ars Technica）↗ ← 返回资讯列表 相关阅读US government website used Chinese model the FBI called "malicious" 今天AI hallucination nearly triggers US military operation 今天AI hallucination of Chinese nuclear components almost led to US military attack 今天FAA tees up $875M AI tool to help manage air traffic congestion 今天A new kind of AI model from a ChatGPT inventor is thrilling developers 今天Security researchers used Anthropic's Claude to hack OpenAI's internal systems in under 72 hours 今天