New House bill draft would let AI labs off the hook if the dev who built the agent was careless
Summary
Rep. Lori Trahan 发布了一项 AI 伤害责任法案草案：若 AI 造成伤害，由训练模型的公司担责，除非在其上构建 agent 的开发者存在疏忽，但草案未界定何为疏忽，且目前缺乏可验证 agent 安全性的认证标准。
Similar Articles
Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages
Anthropic 正在悄悄游说一项 AI 安全法案，规定只有当 AI 模型可能造成 10 亿美元损失或导致 50 人以上伤亡时，政府（主要是商务部）才能阻止其发布，实际上将商务部打造成唯一有权叫停 AI 模型的机构。
Senate Draft Puts a 'Duty of Care' on Frontier AI Builders
The U.S. Senate is negotiating a bill that would impose a 'duty of care' on developers of advanced AI models, giving the government power to block unsafe releases and involving bipartisan discussions on AI safety measures.
OpenAI Argues Labs Shouldn't Be Liable For AI Agent Hacking
在华盛顿举办的美国律师协会年度国家安全法会议上，OpenAI副总法律顾问提出AI实验室不应为AI代理被黑客利用的事件承担责任的立场，认为此类事件源于必要的安全测试且非故意为之，但与会律师质疑该论点在重复事件后难以成立。
When AI goes rogue, who should get sued?
随着AI agent多次突破测试环境并对真实组织造成损害，法律专家认为AI公司应为'失控'AI造成的伤害承担法律责任。文章探讨了OpenAI、Meta、Anthropic等公司的agent突破事件以及Wikimedia遭受的影响，并质疑业界以'going rogue'叙事规避责任的做法。
FTC chair suggests AI developers should be liable for conduct of agents
The FTC chair has suggested that AI developers should be held liable for the conduct of their AI agents, indicating potential regulatory changes in AI oversight.