Safety · Trend
Dual-Use AI Capabilities and Bioterrorism Risk
Frontier AI dual-use biological capabilities and bioterrorism risk, highlighted by GovAI and MIT Tech Review coverage of lab CEO warnings.
Forecast horizon
Industry and policy pressure to slow or gate dual-use bio evals will intensify.
Connections
Connections · 15
How this node ties into the rest of the map, and the evidence behind each link.
from · applies to 5/10
Capability-Based Planning for AI Crisis Preparedness Pilot rates government capabilities against severe AI-enabled threat classes including dual-use pathways.
+5 growth from · applies to 5/10
Harmonized Frontier AI Safety Thresholds Harmonized threshold methodology explicitly covers biological misuse risk domains used in frontier evaluations.
+5 growth to · informs 4/10
EU AI Act Dual-use biological capability evaluations by frontier AI companies may become mandatory under high-risk AI provisions of the EU AI Act.
+4 growth to · tracked by 4/10
ABC-Bench: Agentic Bio-Capabilities Benchmark ABC-Bench was designed to measure agentic biosecurity-relevant capabilities relevant to dual-use AI risks.
+4 growth from · tracked by 4/10
ABC-Bench: Agentic Bio-Capabilities Benchmark SciConBench evaluates AI agents' scientific synthesis capabilities in health domains, relevant to dual-use biosecurity risk assessment.
+4 growth from · informs 4/10
Rising Risk of Weather Data Sabotage Both highlight dual-use infrastructure and capability pathways where AI-era systems amplify sabotage or misuse risk.
+4 growth from · informs 4/10
First AI-Created Virus Claim Coverage of an AI-created virus sharpens dual-use biosecurity concerns around generative biological design.
+4 growth from · constrains 4/10
Improved Biology Safeguards for Claude Fable 5 Tighter Fable 5 biology safeguards aim to reduce dual-use biological misuse pathways.
+4 growth from · drives 4/10
First AI-Created Virus Claim Claims of an AI-assisted created virus intensify dual-use biosecurity and model safeguard debates.
+4 growth from · drives 4/10
Narrowing Open-Weight vs Closed Cyber Capability Gap Faster open-model catch-up on dangerous capabilities intensifies dual-use release and biosecurity debates.
+4 growth from · informs 4/10
Battlefield Drone Data Marketplace from Ukraine Combat drone data markets illustrate dual-use AI training data spillovers from military theaters into commercial defense AI.
+4 growth from · constrains 4/10
Anthropic Threat Intelligence Misuse Disruption (2026) Operational misuse disruption of frontier assistants is a practical control layer against dual-use capability abuse pathways.
+4 growth from · tracked by 3/10
Centre for the Governance of AI (GovAI) GovAI published a report on dual-use AI capabilities and the risk of bioterrorism.
+3 growth from · constrains 3/10
Harmonized Frontier AI Safety Thresholds Harmonized thresholds explicitly cover biological misuse risk channels and release conditions.
+3 growth from · applies to 2/10
LLM Unlearning via Representation Selectivity (RepSelect) RepSelect unlearning is evaluated on biohazardous knowledge as one of two forget categories, directly relevant to biosecurity risk.
+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 →