Anthropic Is Having a Month
7 hours ago
- #data leak
- #software development
- #AI security
- Anthropic accidentally exposed nearly 2,000 source code files and 512,000 lines of code for Claude Code in a software update, revealing the product's full architectural blueprint.
- This is the second such incident in a week; last week, Fortune reported that Anthropic unintentionally made nearly 3,000 internal files public, including details on an unannounced AI model.
- Claude Code is a significant command-line tool that helps developers write and edit code using AI, and its success has prompted competitive responses, such as OpenAI refocusing on developers.
- The leak involved the software scaffolding around the AI model—instructions on behavior, tools, and limits—not the model itself, with developers praising it as a production-grade system.
- Anthropic downplayed the incident as a human error in release packaging, not a security breach, though internal reactions were likely more intense, raising concerns about employee accountability.