Safety · Concept
Narrowing Open-Weight vs Closed Cyber Capability Gap
AISI finds leading open models trail frontier closed cyber capability by only 4–7 months.
Forecast horizon
Shortening lag complicates export-control and open-release risk calculus.
Connections
Connections · 3
How this node ties into the rest of the map, and the evidence behind each link.
from · publishes 4/10
UK AI Safety Institute AISI evaluation finds open-weight cyber capabilities only 4–7 months behind frontier closed models.
+4 growth to · drives 4/10
Dual-Use AI Capabilities and Bioterrorism Risk Faster open-model catch-up on dangerous capabilities intensifies dual-use release and biosecurity debates.
+4 growth to · constrains 3/10
Responsible Open Source and Openness in AI Faster open-weight cyber catch-up pressures open-release governance and evaluation norms.
+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 →