LUCAMEMicrosoft has found a malicious browser or editor extension pretending to be Claude Code, Anthropic's AI coding assistant. Once installed, it quietly sets up a persistent SSH backdoor, meaning attackers can get back into the machine even after a reboot or a password change. It's the kind of threat that's particularly nasty because it looks legitimate at first glance, and developers are often the ones installing these tools with elevated permissions.
What this means for your business
If anyone in your business uses AI coding assistants or installs extensions from outside official marketplaces, this is worth a conversation. The practical step is straightforward: check that any Claude Code tooling comes directly from Anthropic's official channels, and make sure your endpoint security is monitoring for unexpected SSH processes or outbound connections. If you're not sure what's running on developer machines, now's a good time to find out.