Hi,
It's been pretty crazy: Anthropic leaked Claude Code source code.
However this happened - I'm sure some people explained this well, it seems crazy that the hottest AI company right now made such a mistake. Claude Code is the most hyped product at the moment - driving huge acquisition. At first, we can even think that they leaked it on purpose - but I don't think this is the case.
Ironically, developers at Anthropic proudly claimed that Claude Code was writing 100% of its own code...
Mistake is human, but it seems mistake is AI too.
Anthropic sent DMCA notices to individuals who were copying or reproducing Claude Code source code. It could happen that a few people cloned it locally before it was removed.
Now: what's the implication?
For the source code itself: much less than it looks like. Sure - everybody can see how Anthropic engineers built production-ready AI agents. The problematic part? They also leaked Claude Code's roadmap into .md files. Concurrent like OpenAI or OpenCode can take advantage of that. But Claude Code wouldn't be that popular if not using the Anthropic models.
On the business side - nearly no one switched from their Claude Code instance to a local one using source code. Why? Because Anthropic engineers ship so hard that you better use the latest version to get the latest features - and also because you have access to Anthropic AI models with a subscription - and so fixed costs, and ultimately it costs less than using the pay-per-use API key (they burn a lot of cash!). At some extent, it even makes people speak about it - and might bring new customers.
There's something else that made less noise but is much more critical: Mercor, the startup that provides top-tier experts data to train LLMs (OpenAI & Anthropic are customers) has a major data breach. All the exclusive data they had can now be used by concurrents Labs, especially Chinese Labs, making it a national security threat.
In a week or so, everybody will have forgotten about Claude Code leak when they'll release a new model. However, Chinese open-source models is what terrifies Anthropic as they can kill its business.