benchmark

Cyber Weapon Index

Booz Allen ranking of models as autonomous attackers against a live network, covering nine American and nine Chinese models.

Current clusters

product1 distinct publisher

An attack harness closed 67 points of Booz Allen's own AI threat ranking

Booz Allen scored 18 frontier models on a live intrusion and placed Claude Sonnet 5 fifteenth, then paired it with an attack harness and watched it rival the winner. The result: anyone tiering risk by model name is reading a column that measures the wrong object.

Publishers:thenextweb.com

Reality

Evidence52
Adoption25
Hype gap+15
Incentives78

product1 distinct publisher

CrowdStrike will police the OpenAI agents it also puts to work

CrowdStrike will run OpenAI's cyber-tuned model inside its own harness while policing OpenAI's Codex agents at runtime, which leaves one vendor configuring what agents can reach and auditing what they did.

Publishers:thenextweb.com

Reality

Evidence46
Adoption24