Anthropic puts $35M behind open-source cyber defense

Anthropic is putting $35 million in Claude credits into its Defender Advantage Fund, while making Claude Mythos 5 available for vulnerability scans in Claude Security, according to Anthropic’s defender announcement. The model returns findings, severity and confidence ratings, and suggested patches rather than giving users direct access to its cyber capabilities; every patch still requires human approval. Security teams should file the distinction beside the archive’s runtime-security analysis: frontier cyber capability is becoming a governed service output, not an unrestricted prompt box.