Study of AI-Assisted Cyberattacks May Reshape How Security Industry Measures Risk

**Anthropic analysis of 832 banned accounts shows AI is increasingly used for post-compromise activity (account discovery, lateral movement, privilege escalation) and for writing malware, raising operational risk as AI enables less-skilled actors to perform sophisticated techniques and enabling agentic attack orchestration that current frameworks like MITRE ATT&CK do not capture.**