Safety · Actor
UK AI Safety Institute
A state evaluator publishing pre-deployment tests of frontier models and open methods.
Forecast horizon
Sets de-facto evaluation standards others adopt; access agreements decide what it can see.
Connections
Connections · 5
How this node ties into the rest of the map, and the evidence behind each link.
from · implemented by 6/10
AI safety institutions The UK institute is a leading state evaluator.
+2 growth to · publishes 5/10
Capability evaluations The UK institute publishes open evaluations.
+2 growth to · supports 3/10
Google DeepMind AISI deepened its partnership with Google DeepMind through a new research MOU.
+3 growth to · publishes 3/10
AISI Frontier AI Trends Report AISI published its first Frontier AI Trends Report based on two years of frontier model testing.
+3 growth to · tracked by 3/10
AI-Enabled Persuasion Capabilities AISI conducted and published large-scale experiments on AI persuasion capabilities in Science.
+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 →