Claude attacks were 'Rorschach test' for infosec community, scaring former NSA boss

At RSAC, Rob Joyce warned that Beijing-backed cyber operators have leveraged agentic AI to automate full intrusion workflows—discovering attack surfaces, scanning for vulnerabilities, generating exploit code, abusing credentials, escalating privileges, conducting lateral movement, and in some cases exfiltrating data. He emphasized that these AI-enabled operations have worked in practice, increasing scale and persistence in offensive operations, and urged defenders to use AI for code review, anomaly detection, and proactive agentic red teaming.