The National Academies of Sciences, Engineering, and Medicine will, as directed by the President in Executive Order 14110, conduct a consensus study that will:
(A) Assess the ways in which artificial intelligence (AI) can increase biosecurity risks, including risks from generative AI trained on biological data, and make recommendations on how to mitigate those risks;
(B) Consider the national security implications of the use of data and datasets, especially those associated with pathogens and omics studies, that the United States Government hosts, generates, funds the creation of, or otherwise owns, for the training of generative AI models, and make recommendations on how to mitigate the risks related to the use of these data and datasets; and
(C) Assess the ways in which AI applied to biology can be used to reduce biosecurity risks, including recommendations on opportunities to coordinate data and high-performance computing resources.

This study will focus specifically on concerns of transmissible biological threats that could pose significant epidemic and pandemic-scale consequences. This study will build on the 2018 study Biodefense in an Age of Synthetic Biology but would not duplicate that work. An unclassified consensus report will provide evidence-supported analysis of AI model capabilities, including trends and actionable recommendations for near- and mid-term implementation.