Anthropic has cultivated a reputation as the conscientious AI firm, emphasizing careful development. The company releases extensive studies on AI safety, hires top experts in the field, and frequently discusses the ethical duties involved in creating advanced technology—so much so that it is currently engaged in a dispute with the Department of Defense. However, on Tuesday, an oversight occurred when someone neglected a crucial step. Notably, this is the second such incident within a week.

Last Thursday, it was reported that Anthropic unintentionally made nearly 3,000 internal documents accessible to the public, which included a draft announcement for an unlaunched, advanced model. Then, on Tuesday, when Anthropic updated its Claude Code software package to version 2.1.88, a file was mistakenly included that revealed almost 2,000 source code files and over 512,000 lines of code—effectively laying bare the complete structural design of one of its key products. Security researcher Chaofan Shou spotted the issue almost instantly and shared it on X.

Anthropic’s response to various media outlets was relatively calm, stating: “This was a release packaging issue caused by human error, not a security breach.” Internally, however, the reaction was likely far more intense.

Claude Code is a significant offering, not a minor one. It is a command-line tool that enables developers to utilize Anthropic’s AI for writing and editing code, and it has grown robust enough to concern competitors. Reports indicate that OpenAI discontinued its public video generation tool Sora just six months after launch to reorient toward developers and enterprises—partly due to Claude Code’s increasing influence.

The leak did not involve the AI model itself but rather the surrounding software framework—the guidelines that dictate the model’s behavior, tool usage, and boundaries. Developers quickly published in-depth evaluations, with one noting that the product represents “a production-grade developer experience, not just a wrapper around an API.”

Whether this incident has any lasting impact is a question for developers to determine. Rivals might gain insights from the exposed architecture, though the industry evolves rapidly. Regardless, one can imagine a highly skilled engineer at Anthropic spending the day anxiously pondering their job security. Hopefully, it is not the same individual or team responsible for last week’s oversight.