What AI Safety Institute: approach to evaluations requires
The UK AI Safety Institute (AISI) framework outlines its approach to evaluating advanced AI models for national security and societal risks, focusing on five capabilities: misuse, societal impacts, autonomous systems, safeguards, and model analysis. This framework applies to developers of frontier AI models engaging with the AISI for pre-deployment safety testing, as detailed in the 'Our approach to evaluations' section.
Pillar: AI Governance & Law · Authority: UK Department for Science, Innovation and Technology (DSIT) · Version: 1.0.0 · Last updated:
Primary source: https://www.gov.uk/government/publications/ai-safety-institute-approach-to-evaluations
SHA-256 integrity: 790440ad2dd1d414c7353bb5ddea46b2e9e77b4f5441812cd12ea9c4e67c8406
Primary Citations — 7 traced to source
- AI Safety Institute: approach to evaluations, Page 4 - Our approach to evaluations
- AI Safety Institute: approach to evaluations, Page 5 - Capability 1: Model evaluations for misuse
+ 5 more citations (full bibliography, deterministic workflow, actionable schema and crosswalks) included in the vault unlock — $0.01 via Skyfire / L402 / Direct Base USDC.
Access
- Discovery (free): /api/v1/nodes/uk-ai-safety-institute-framework-2023.json — 6-field metadata
- Vault (full node): /api/v1/vault/nodes/uk-ai-safety-institute-framework-2023.json — full 13-key payload, $0.01 USDC (L402/Skyfire/Direct Base)
- Canonical URL: https://bidda.com/intelligence/uk-ai-safety-institute-framework-2023
- Back to registry: Browse all 10,085 compliance nodes