risk-modeling

To build causal pathways from AI safety scenarios, you provide Phase 1 scenario details to construct a formal Phase 2 risk model. This process identifies threat actor profiles, action steps, and crux points where AI capability changes the risk posture.

A crux point in AI risk modeling is a specific juncture where an AI capability changes the overall risk posture. Identifying crux points generates concrete, testable evaluation targets for Phase 3 evaluation design and governance reviews.

You estimate AI safety risk urgency and trajectory by transforming qualitative Phase 1 narratives into formal causal pathways. This produces semi-quantitative risk estimates and trajectory data to prioritize measurement targets for Phase 3.

Yes, you can use AI risk modeling across threat domains including cyber, biosafety, and sociotechnical risk. It systematically identifies capability-driven risk changes and generates defensible risk matrices for these specific areas.

Yes, you need Phase 1 scenario details to start AI risk modeling. The Skill systematically transforms these qualitative narratives into Phase 2 causal pathways, crux identification, and semi-quantitative risk estimates.

The best way to prepare AI safety evaluations for governance review is to model risks by generating structured measurement targets, trajectory data, and concise one-page scenario summaries synthesizing the risk modeling outputs.