Anthropic scrambles after Claude Code AI source code leaked online
Anthropic, the team behind Claude Code AI, is scrambling after its source code leaked online.
While no customer data or core AI secrets were exposed, the leak did reveal important details about how their AI works, raising fresh questions about how Anthropic handles copyright.
Anthropic trims takedowns amid copyright scrutiny
Anthropic first tried to remove more than 8,000 copies of the leaked code but later trimmed that list to just 96 after realizing they'd cast too wide a net.
The leak happened because a leftover file in their software accidentally pointed people to the code.
This isn't Anthropic's first copyright mess: they previously settled for $1.5 billion over using pirated books and ran Project Panama to secretly process used books for training data.
All this puts a spotlight on transparency and IP rights in the fast-moving AI world.