Anthropic accidentally takes down thousands of GitHub repositories while trying to contain a code leak

Anthropic is at the center of a high-profile controversy: while trying to remove leaked source code from GitHub, the company mistakenly sent thousands of copyright infringement notices, forcing the takedown of a huge number of repositories that had nothing to do with the leak. It became one of the largest mass-DMCA incidents in the platform's history.

The company was following a standard intellectual property protection routine: it discovered that its proprietary code had appeared in public and tried to remove it quickly. But the automated notice-filing process got out of hand. The fallout hit not only repositories containing the leaked material but also thousands of entirely unrelated projects.

Anthropic's leadership acknowledged the error and said it was unintentional. The company retracted most of the notices it had sent and is working with GitHub to undo the damage for affected users. For developers whose repositories became unavailable, it meant losing access to their own projects, even if only temporarily.

The leak itself deserves separate attention. The company develops the Claude family of models, among the most sought-after in the enterprise segment, and has traditionally kept its technical work under tight secrecy. Source code appearing in public points to a serious failure in information security. The response, a wave of mass DMCA notices, looks like an attempt to stop the spread of the material as fast as possible, without first assessing the scale of collateral damage.

The situation exposes a systemic problem: automated copyright enforcement tools are fast, but not always accurate. Algorithms that search for similar code can wrongly flag legitimate projects as infringers, especially if the leaked fragment has spread widely across forks, mirrors and tutorial repositories. When such tools are applied at scale without sufficient verification, any flaw in the selection criteria turns into a mass failure.

GitHub found itself in a dual position. Under US law, the platform must respond to DMCA requests from rights holders, and it has no technical means of checking every request for correctness in real time. The largest developer platform became an unwitting instrument of someone else's mistake.

The incident raises a broader question about how technology companies should approach intellectual property protection in the era of open platforms. The "block now, sort it out later" principle is tactically understandable, but in practice it causes collateral damage. The people who get hurt are the very developers and researchers who build trust in the company and use its products.

Retracting the notices fixes the immediate consequences, but it does not settle the reputational questions. Anthropic actively positions itself as a responsible player in the AI industry, with an emphasis on safety and trust. An episode of mass removal of other people's repositories, even an accidental one, conflicts with that image and shows that even companies with high standards can see internal processes fail at the worst possible moment.