2025 Workshop on the Nexus of Biosecurity, AI, and Modern Conflict: A Tabletop Exercise

Description

This workshop convened a diverse group of interdisciplinary experts to explore the vulnerabilities and risks posed by the convergence of artificial intelligence (AI) and biotechnology—particularly in the context of modern conflict. Building on the foundation laid by an inaugural workshop in April 2024, this second gathering shifted from exploration to action through a dynamic, scenario-based tabletop exercise designed to anticipate and address emerging biosecurity threats.

The inaugural workshop explored critical questions surrounding the future of AI-enabled biotechnology, including its potential to accelerate medical countermeasure (MCM) design, the dual-use dilemma of AI's protein engineering capabilities, the evolving role of AI in cybersecurity, and the crucial lessons learned from securing cyber-physical systems. These discussions, along with insights gleaned from academic literature [1] and key reports [2], underscored the urgent need for the development of systematic threat models for the biomedical R&D and biomanufacturing sectors. This second workshop directly responds to that need.

The tabletop exercise simulated real-world vulnerabilities in a notional biomanufacturing or biosurveillance system constructed using publicly available data. Participants—drawn from biology, cybersecurity, AI, and systems engineering domains—adopted an adversarial lens to assess how AI integration across discovery, production, and testing layers could be exploited by malicious actors. Importantly, the exercise extended beyond the threat of engineered biological agents to consider broader system-level risks, including supply chain compromises, data poisoning, and infrastructure attacks.

Participants engaged in four structured breakout sessions:

Biomanufacturing/Biosurveillance System Characterization: Mapped a detailed model of a representative biomanufacturing or biosurveillance pipeline, including AI technologies, cyber-physical components, and critical dependencies. This foundational understanding is crucial for subsequent vulnerability analysis.
AI Technology Vulnerabilities: Explored threat vectors such as adversarial inputs, training data manipulation, model inversion, and AI hallucinations.
Cyber and Cyber-Physical Infrastructure Vulnerabilities: Investigated plausible attack pathways targeting infrastructure components and assessed the impact of disruptions on medical countermeasure availability, public trust, and supply chain integrity.
Countermeasures and Mitigations: Identified and evaluated a range of potential defenses, from zero-trust architectures and anomaly detection systems to process and policy interventions.

Each session began with a framing presentation followed by in-depth group discussion and analysis. The workshop fostered a shared vocabulary and built momentum toward a coordinated research and policy agenda to safeguard the bioeconomy.

The findings and recommendations from this tabletop exercise, together with insights from the April 2024 workshop, will be synthesized in a comprehensive joint report. This report will capture the evolving conversation around AI-enabled biotechnology, detailing the interdisciplinary threat models, identified vulnerabilities, and mitigation strategies explored across both events. By integrating perspectives from researchers, practitioners, and policymakers, the report will help guide future efforts to strengthen biosecurity in an era defined by rapid technological convergence and increasing geopolitical complexity.

References

[1] Elgabry M, Johnson S. Cyber-biological convergence: a systematic review and future outlook. Frontiers in Bioengineering and Biotechnology. 2024 Sep 24;12:1456354. https://pmc.ncbi.nlm.nih.gov/articles/PMC11458441/  

[2] National Academies of Sciences, Engineering, and Medicine. 2025. The Age of AI in the Life Sciences: Benefits and Biosecurity Considerations. Washington, DC: The National Academies Press. https://doi.org/10.17226/28868