A malicious actor compromised a third-party extension for the Cursor code editor that connects to Anthropic's Claude AI, injecting harmful code into the tool developers use to write software. Anthropic says its own systems weren't breached, but the attack shows how AI-assisted development tools can become a route into a business's code.
What this means for your business
If anyone in your business uses Cursor or Claude Code extensions to write or review code, check which extensions are installed and remove anything you don't recognise. Supply-chain attacks like this are designed to be quiet, so the risk isn't just to the tool itself but to whatever code was produced or reviewed while the compromised extension was running. It's worth treating any code written during that period with extra scrutiny before it goes anywhere near a production system.
LUCAME