— Creation of biological weapons is not a new concept or risk, but the potential for AI-enabled biological tools to affect this risk is a new paradigm.
— AI tools to predict a protein’s three-dimensional structure, important for understanding mechanisms of disease and developing new countermeasures, have advanced substantially in the past few years.
— For drug-discovery applications, AI tools can generate thousands of molecules as potential drug candidates in days, a feat that could take years without AI methods.
— Data and research studies evaluating the biosecurity risks of AI-enabled biological tools are lacking.
A new report produced by an expert panel at the National Academies of Sciences, Engineering, and Medicine assesses how AI-enabled biological tools could uniquely impact biosecurity risk, and how advancements in such tools could also be used to mitigate these risks.
AI-enabled biological tools contribute to multiple aspects of biodefense, from improving infectious disease surveillance by epidemiological tracking and forecasting, and accelerating the design of countermeasures, such as diagnostic tests or medications. Research programs that produce knowledge about and build datasets on the biology of infectious agents would help support the development of such countermeasures.
The report sets out recommendations for continuing support of research programs utilizing AI tools to better understand the biology of infectious agents, conduct biosurveillance programs, and develop medical countermeasures.
Highlighted recommendations include:
- Federal agencies should provide strong incentives and infrastructure for the standardization, curation, integration, and continuous maintenance of high-quality, publicly accessible biological data at scale.
- Support funding efforts and infrastructures that could create and steward AI-compatible data and training sets for biological applications.
- More research in new methodologies for nucleic acid synthesis screening, including how to leverage AI–enabled biological tools for screening, is needed to be an efficient control point and possible strategy to mitigate potential biosecurity risks.
- Federal agencies that house or fund biological databases should consider them to be strategic assets and concordantly implement robust data provenance mechanisms to maintain the highest data quality.
- U.S. Department of Defense and the U.S. AI Safety Institute should develop an If-Then strategy to continuously evaluate both the availability and quality of data, and the associated AI-enabled capabilities that emerge, to anticipate changes in the risk landscape.
- If clear associations between viral sequences and virulence parameters become known, then evaluate the capability of AI models to predict or design pathogenicity and/or virulence.
- If robust viral phylogenomic sequence datasets linked to epidemiological data become available, then assess for the development of new AI models of transmissibility that could be used to design new threats.
- Engaging practitioners and developers of AI-enabled biological tools to explore the most effective approaches for implementation is important, as is engaging biologists and other developers to assess impacts of screening on public health preparedness and response.
- To maximize the benefits of AI in biology, U.S. federal research funding agencies such as the National Institutes of Health, Advanced Research Projects Agency for Health, Defense Advanced Research Projects Agency, National Science Foundation, U.S. Department of Energy, and others should fund efforts and infrastructures to steward AI-compatible data and training sets for biological applications that are made publicly available and support public–private partnerships for the generation of training datasets.
- While a public–private partnership is a reasonable approach to pilot the creation of an AI–ready data resource on government-owned infrastructure, a dedicated organization is central to a sustainable national data stewardship resource. The National Artificial Intelligence Research Resource (NAIRR) pilot should explore approaches for establishing central repositories for training AI models.
The assessment was requested by the Department of Defense.
READ THE FULL REPORT:
The Age of AI in the Life Sciences: Benefits and Biosecurity Considerations (2025). Committee on Assessing and Navigating Biosecurity Concerns and Benefits of Artificial Intelligence Use in the Life Sciences.