在2026年3月31日,Anthropic意外暴露了其AI编码工具的内部代码. Anthropic accidentally exposed internal code for its AI coding tool on March 31, 2026.

由于包装错误,Anthropic 在 2026 年 3 月 31 日意外暴露了其 AI 编码工具 Claude Code 的内部源代码. Anthropic accidentally exposed the internal source code for its AI coding tool, Claude Code, due to a packaging error on March 31, 2026. 泄露发生在一个错误地发布到npm软件注册表的调试文件中, The leak occurred when a debugging file was mistakenly published on the npm software registry, revealing 512,000 lines of code. 虽然没有客户数据或模型被曝光,但泄露给竞争对手提供了该工具功能和高级功能的蓝图. While no customer data or models were exposed, the leak provided competitors with a blueprint for the tool's functionality and advanced features. 公司确认此事是人为错误, 正在努力防止未来的事件. Anthropic confirmed the issue was a human error and is working to prevent future incidents.