Anthropic is having a month

Anthropic has built its public identity around the successful idea of ​​being a mindful AI company. We have published detailed research on AI risks, hired the best researchers in the field, and spoken out about the responsibility that comes with building these powerful technologies. Of course, the voices are so loud that they are now fighting against it with the Ministry of Defense. Unfortunately on Tuesday someone there forgot to check the box.

Especially the second time in a week. Last Thursday, Fortune reported that Anthropic had inadvertently released nearly 3,000 internal files to the public, including draft blog posts describing a powerful new model that the company had not yet announced.

Here’s what happened Tuesday: When Anthropic released version 2.1.88 of the Claude Code software package, Anthropic inadvertently included files that exposed nearly 2,000 source code files and more than 512,000 lines of code. This is essentially a complete architectural blueprint for one of our most important products. A security researcher named Chaofan Shou noticed this almost immediately and posted it on X. Anthropic’s statement to several media outlets was nonchalant: “This was not a security breach, but a release packaging issue caused by human error.” (I’m guessing things were less measured internally.)

Claude Code is not a trivial product. It’s a command-line tool that lets developers write and edit code using Anthropic’s AI, and it’s become powerful enough to make competitors nervous. According to WSJ, OpenAI has discontinued its video creation product, Sora, just six months after launching it to the public in order to refocus its efforts on developers and enterprises. In part, it was a response to Claude Code’s growth momentum.

What was leaked wasn’t the AI ​​model itself, but the software scaffolding around it – the instructions that tell it how to behave, what tools to use, and where its limits are. Developers began publishing detailed analyzes almost immediately, describing the product as “a production-grade developer experience rather than just an API wrapper.”

Whether this matters in any lasting way is a question best suited to developers. Competitors may find this architecture beneficial. At the same time, the field moves quickly.

Either way, you can imagine that somewhere at Anthropic, a very talented engineer spent the rest of his days quietly wondering if they still had a job. I just hope it’s not the same engineers or engineering team as at the end of last week.