What We Learned Mapping a Year's Worth of AI-Enabled Cyber Threats

A verbatim reading of two complementary Anthropic reports on AI-enabled cyber threats. The first is the overview published on anthropic.com, covering three headline conclusions from a year-long study of 832 banned accounts. The second is the full technical analysis from the Frontier Red Team blog, covering the ARiES risk-scoring methodology, a detailed breakdown of how attackers use AI across the kill chain, the GTG-1002 agentic espionage case study, and the case for expanding the MITRE ATT&CK framework to capture autonomous AI-orchestrated attacks.