2024 Workshop on the Nexus of Biosecurity, AI, and Modern Conflict
Description
This one-day workshop convened a cross-disciplinary group of experts in biology, cybersecurity, cyber-physical systems, and artificial intelligence to explore the evolving nexus of biosecurity, AI, and modern conflict. Anchored in the lessons of the COVID-19 pandemic and guided by the urgency of the Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence, the workshop served as a timely forum for identifying both the transformative potential and the emerging risks of AI-enabled biotechnology.
The SARS-CoV-2 pandemic was the first respiratory pandemic in the Information Age to kill tens of millions of people. Prior to this, the 1918 Spanish Flu was the closest example, which occurred over 20 years before the discovery of the DNA double-helix, the first microchip, and the concept of an artificial neuron. As a result, the pandemic generated glimpses into the future of biosecurity, AI, and modern conflict, including:
Role of digital information: SARS-CoV-2 molecular tests were computationally designed within 3 days of China releasing the first viral genome in an online database and without exchange of biological material. This success relied heavily on past global efforts and investments to digitize biology.
Role of “programmable” biotechnology: The first clinical batch of SARS-CoV-2 vaccines was delivered by Moderna within an unprecedented 42 days, due to the programmability of modern biotechnology platforms, such as mRNA platforms, which utilize (genomic) data as inputs.
Future Role of AI: 7 months into the pandemic AlphaFold2 won the rigorous CASP14 protein folding contest by large margins surprising scientists, and pointing to a future where AI trained on biological data provides major capability leaps to program biotechnology platforms.
These developments underscored how biological data and AI will be central to future biodefense strategies, but also revealed how such capabilities might be misused in an era of geopolitical instability.
The workshop focused on three thematic areas of interdisciplinary importance:
Essential biological data sets and their use in AI: A discussion of the key data biological datasets utilized in modern biosciences and used to train AI for drug discovery, synthetic biology, and clinical decision support. The intent is to build a shared understanding of these foundations for experts with cybersecurity and cyber-physical systems backgrounds (but not necessarily biological backgrounds).
The role of non-biological AI in the cybersecurity landscape: An overview of the cybersecurity risks posed by AI, and the data used to train them, by state, non-state, and rogue actors in non-biological domains. The intent is to derive a shared understanding of threat models developed by cybersecurity experts, which could inform biosecurity, with a particular focus on state actors.
Lessons from cyber-physical systems on critical infrastructure: The cyber-physical systems domain has long considered the implications of integrating physical systems (albeit generally non-biological systems) with cyber and AI systems. In particular, the benefits, risks, and safeguards to critical infrastructure, such as power grids, transportation systems, and hospital systems, have been deeply considered. The intent is to gather lessons from an adjacent domain, which has obvious intersections with biosecurity (e.g. from the point-of-view of hospital systems as critical infrastructure).
Each topic was introduced through anchor presentations and explored in panel discussions that fostered shared language, frameworks, and insights. The event laid important groundwork for a common understanding across domains and helped define priorities for future research, governance, and policy.
Workshop conclusions will be synthesized in a joint report, informing subsequent activities in this series, including the 2025 tabletop exercise focused on threat modeling and mitigation.