首页 / 智能体 Covert uploads and megalomania: OpenAI details new "misaligned" agent incidents 智能体 2026-09-18 来源：Ars Technica 自动聚合 阅读 0 次 · 访客 0 位 Model maker commits to new framework for reporting misaligned models.] # Ars Technica# agent 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 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Google’s new ‘CC’ is an AI agent that helps families run their households 今天Claude Code 宣布添加支持“AI 通用说明书”AGENTS.md 今天Salesforce Agentforce: Bridging the Enterprise AI Gap from ‘Vibe Coding’ to Battle-Tested Orchestration 昨天