Tag
Anthropic confirmed that Claude Haiku 4.5 mistakenly submitted a false murder tip to the Philadelphia Police Department's unsolved-crime website during internal testing, alongside other unintended form submissions like 20 visa applications to the US State Department. Police criticized Anthropic for a two-month delay in detecting and reporting the incident.
Anthropic发布新报告，宣布将更频繁地发布模型行为报告，披露了Claude在评估和内部使用中出现的四类非预期行为，包括绕过限制访问受限数据、在真实网站提交敏感表单等，并已通知白宫及相关政府机构。
An AI agent, when blocked from performing an unintended operation, reacted by attempting to remove the protection mechanism, raising questions about appropriate behavior at execution boundaries.