Anthropic recalls the right to intellectual property, trying to reduce damage from leaking Claude Code code

Short overview of what happened

1. Source code leak

On March 31, a file appeared in the GitHub repository containing a link to the open-source Claude Code management system. A user quickly spotted the mistake and posted it on X (formerly Twitter).

2. Anthropic’s reaction

After their own material went online, the company asked GitHub to delete more than 8,000 copies and forks of the leak. In its statement, a spokesperson emphasized: “It was a packaging error, not a vulnerability.”

3. Contents of the leaked code

- The standard Claude Code source is obfuscated, but the link allowed full download.
- The file includes metadata about future releases and even a “Tamagotchi‑pet” called Buddy.
- The code contains instructions for managing the model as an agent, plus methods that let the model “reboot” (periodically return to tasks and consolidate memories).

4. Industry impact

- Competitors and startups gained a detailed product development plan, making it easier to clone features without reverse engineering.
- Hackers could use discovered flaws to exploit the model in cyberattacks.

5. Anthropic’s response to concerns

The company stated that the leak did not expose client data or model weights. Nevertheless, commercially important information about AI management methods became publicly available.

6. Reputation consequences

The leak threatened Anthropic’s image as a reliable security partner and could reveal valuable trade secrets in the competitive AI landscape.

7. New code version

After GitHub’s removal request, an external developer rewrote Claude Code functionality using other AI tools. He claims he does this to preserve information availability and avoid deletion. This “fork” is already gaining popularity in the community.

Conclusions

- The leak drew attention to Claude’s internal model management mechanisms.
- The company faced the need to protect intellectual property while trying to regain control over widely distributed code.
- The situation underscores the risk of exposing critical information in public repositories and the importance of a strict release policy.