Safety · Concept
Threat-Model Coverage Gap in AI Safety Evaluation
Methodological critique that automated red-teaming gains on fixed benchmarks do not replace human evaluators with different deployment-context coverage.
Forecast horizon
Pushes evaluation science toward external, context-divergent evaluators rather than benchmark-only automation.
Connections
Connections · 4
How this node ties into the rest of the map, and the evidence behind each link.
to · informs 3/10
Open Problems in Technical AI Risk Modeling Coverage-gap critique shows benchmark-internal validity can miss deployment harms, feeding open risk-modeling questions.
+3 growth to · constrains 3/10
Science of AI Evaluation Coverage-gap argument limits claims that cheaper automated red-teaming can retire human/contextual evaluators.
+3 growth to · informs 3/10
Science of AI Evaluation Defines threat-model coverage gap showing benchmark red-teaming completeness is not deployment-context coverage.
+3 growth from · publishes 3/10
arXiv AI Safety Feed arXiv AI safety item posts the translational note defining the threat-model coverage gap.
+3 growthSignal sources
Signal sources
Dated facts from primary sources in this direction.
US evaluation centre Jun 2025
In June 2025 the US AI Safety Institute was renamed the Center for AI Standards and Innovation (CAISI), pivoting toward security, standards and adversary-model assessment.
NIST → Frontier safeguards May 2025
Anthropic activated its ASL-3 deployment and security standard with Claude Opus 4 on 22 May 2025 — the first real-world trigger of a responsible-scaling tier, focused on blocking bio-weapon uplift.
Anthropic → Cross-border testing 2025
The International Network of AI Safety Institutes (launched Nov 2024) ran a third joint testing exercise focused on agentic AI systems across cyber and fraud strands.
European Commission — AI Office →