Safety · Concept
EvalSafetyGap: LLM Evaluation-Safety Measurement Framework
Hybrid survey and conceptual framework identifying gaps between benchmark scores and latent safety properties in LLMs.
Forecast horizon
Connections
Connections · 6
How this node ties into the rest of the map, and the evidence behind each link.
from · informs 3/10
Industry-Wide Jailbreak Severity Scoring Framework Industry jailbreak severity scoring addresses the evaluation-safety gap identified in EvalSafetyGap framework.
+3 growth to · references 2/10
Science of AI Evaluation EvalSafetyGap framework contributes to the science of AI evaluation by formalizing the gap between evaluation proxies and alignment properties.
+2 growth to · supports 2/10
Science of AI Evaluation EvalSafetyGap contributes to the science of AI evaluation by identifying systematic gaps between benchmark scores and latent safety properties.
+2 growth from · informs 2/10
Adversarial Pragmatics Benchmark for AI Safety Evaluation Adversarial pragmatics benchmarks address the evaluation-safety gap by providing linguistically controlled safety evaluation protocols.
+2 growth from · frames 2/10
Strategic Red Teaming as Board-Level AI Governance Instrument Strategic red teaming addresses the same governance gap that EvalSafetyGap identifies: proxy metrics improving while latent risks remain unverified.
+2 growth from · informs 2/10
Strategic Red Teaming as Board-Level AI Governance Instrument Strategic red teaming as governance instrument addresses the same evaluation-safety measurement gaps identified in EvalSafetyGap.
+2 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 →