AIBRIEF
The grant targets the legal and practical barriers to information sharing between AI model developers, biosecurity experts, and safety researchers when an AI bioweapon is suspected.
OpenAI is funding the Nuclear Threat Initiative to map the legal and practical barriers to cross-border information sharing when an AI-enabled biothreat is suspected. NTI is one of 14 grant recipients chosen from more than 400 applicants.
The project, described by NTI deputy vice president for global biological policy Hayley Severance, would connect AI model developers, biosecurity experts, and safety researchers across jurisdictions and labs. It is an information-sharing study, not an enforcement mechanism: NTI will chart the gaps that keep those groups from exchanging signals today, not build the pipeline that moves the data.
Researchers have already shown AI can help design and synthesize a virus not found in nature, and Severance told Scientific American that "sophisticated actors" will be able to do more with AI as models improve. The grant joins OpenAI's published biology-preparedness work, including its January 2025 paper on future AI capabilities in biology and its policy ideas for the intelligence age, and a separate reported partnership with US national labs on bioweapons detection.
The amount is undisclosed, and that is the other half of the story. OpenAI is paying a nonprofit to map the legal gaps on a capability it ships itself, and NTI's deliverable is a study, not a defense.