Anthropic's Momentum Builds

Anthropic has cultivated its public image around the core belief that it is the responsible, safety-conscious AI company. It publishes extensive research on AI risks, employs leading experts in the field, and has been outspoken about the ethical duties involved in developing such transformative technology—so outspoken, in fact, that it is currently engaged in a dispute with the Department of Defense. On Tuesday, however, someone there overlooked a simple checkbox.

Notably, this is the second such incident within a week. Last Thursday, Fortune reported that Anthropic had inadvertently made nearly 3,000 internal files publicly accessible, including a draft blog post detailing a powerful, unannounced new model.

Here’s what occurred on Tuesday: When Anthropic released version 2.1.88 of its Claude Code software package, it mistakenly included a file that exposed almost 2,000 source code files and over 512,000 lines of code—effectively the complete architectural blueprint for one of its most critical products. Security researcher Chaofan Shou spotted the issue almost instantly and posted about it on X. Anthropic’s statement to various media outlets was relatively calm, as these things go: “This was a release packaging issue caused by human error, not a security breach.” (Internally, the reaction was likely far less composed.)

Claude Code is no minor offering. It’s a command-line tool that enables developers to use Anthropic’s AI for writing and editing code, and it has grown robust enough to make competitors uneasy. According to The Wall Street Journal, OpenAI shut down its public video generation tool Sora just six months after launch to refocus on developers and enterprise clients—partly in reaction to Claude Code’s rising traction.

The leak did not involve the AI model itself, but rather the surrounding software framework—the instructions that govern the model’s behavior, tool usage, and operational boundaries. Developers quickly began publishing detailed analyses, with one noting the product represents “a production-grade developer experience, not just a simple API wrapper.”

Whether this has any lasting impact is a question best answered by developers. Competitors may find the architecture educational, though the field evolves rapidly.

Either way, you can imagine a highly skilled engineer at Anthropic spending the rest of the day quietly contemplating their job security. One can only hope it’s not the same engineer—or engineering team—responsible for last week’s oversight.

Disrupt 2026: The Entire Tech Ecosystem, Under One Roof

Your next funding round. Your next key hire. Your next breakthrough opportunity. Discover it all at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders converge for three days of 250+ tactical sessions, powerful networking, and industry-defining innovation. Register now to save up to $400.

Save up to $300 or 30% on TechCrunch Founder Summit

Join 1,000+ founders and investors at TechCrunch Founder Summit 2026 for a full day dedicated to growth, execution, and real-world scaling. Learn from founders and investors who have shaped the tech landscape. Connect with peers facing similar growth challenges. Leave with actionable tactics you can implement immediately.

Offer ends March 13.