What happened
The Frontier Model Forum published a new issue brief examining how the combination of increasingly autonomous frontier AI agents and AI-enabled biological tools (biological foundation models, specialized analytical tools, automated lab workflows) may introduce novel biosecurity risks beyond either technology alone. The brief outlines two preliminary categories of concern: lowering barriers to entry for advanced biological capabilities, and raising the ceiling of harm achievable by a malicious actor, and sets out preliminary considerations for evaluation methodology, technical safeguards, and monitoring specific to agentic AI-bio combinations. It is drawn from FMF member-firm discussions and the broader biosafety/biosecurity expert community, following on FMF's existing AI-bio misuse taxonomy work.
Why it matters
As frontier labs increasingly deploy agentic systems alongside specialized biological AI tools, this brief gives risk, safety, and biosecurity teams an early framework for evaluating a threat combination that most existing frontier safety frameworks address only in isolation.
Action needed
Map agentic-AI + bio-tool combinations in your organization's AI risk taxonomy and benchmark internal evaluation practices against FMF's proposed considerations.