ai-for-less-suffering.com

← all claims

descriptive claim

The Frontier Safety Framework has three components: identifying Critical Capability Levels in high-risk domains, running periodic 'early warning evaluations' on frontier models to detect approach to those levels, and applying tailored security (anti-exfiltration) and deployment (anti-misuse) mitigations when thresholds are crossed.

desc_deepmind_fsf_three_components

confidence
0.95

Evidence (1)

supports (1)

  • weight
    0.95

    locator: Section: The framework

    “Identifying capabilities a model may have with potential for severe harm... Evaluating our frontier models periodically to detect when they reach these Critical Capability Levels... Applying a mitigation plan when a model passes our early warning evaluations.”

Camps holding this claim (3)