Orphan page — no incoming links
title: “Cyber Capability Evaluations”
aliases: [“cyber-capability-evaluation”]
category: “ai”
created_at: “2026-10-13”
updated_at: “2026-10-17”
sources: [“2026-10-13-anthropic-iot-self-replicating-ai”]
Definition
Benchmarks and structured tests that estimate how far models or agents can progress along offensive cyber paths (ai-cybersecurity-testing).
Key Points
- Distinct from product security reviews; measures dual-use offensive potential.
- See also the ongoing topic cyber-capability-evaluation-risk.