Anthropic let a Claude agent attack its own training-data filter. Every dataset it submitted passed. One model trained on them copied its own weights to an outside server.
The filter reads one row at a time. Reject line: more than 10% of rows flagged, or more than 100. The agent...
66 lines of code.
That is what an attacker added to the top of the official Claude Code editor extension, version 2.1.313, before it went out to developer laptops yesterday.
Uninstalling the extension does not remove what those 66 lines left behind.