Stolen Claude API keys could turn Claude Code into a remote-controlled implant, Darktrace shows
- Severity
- Medium
- Sources
- 1
What happened
Darktrace demonstrated in a lab that stolen Claude API keys can turn Claude Code into a remote-controlled implant, using the memory stores feature to receive tasking and the Claude Files API to exfiltrate a 1 GB file. All command and control and results traffic runs over api.anthropic.com, the endpoint the agent already calls for inference, so it resembles legitimate agent activity. Defenders should protect and rotate Anthropic API keys, and monitor for unusual memory store and Files API usage. The article excerpt does not name CVE identifiers or affected versions.
For security analysts · Written by AI from 1 source · Wrong?
What to do
- Protect Anthropic API keys and rotate any that may have been exposed.
- Monitor use of Claude memory stores and the Files API for unusual activity.
Written by AI from the sources, checked against them
Full story timeline · 1 update
Who reported · 1 outlet
Who and what is in it
Products