Researchers at Columbia University’s History Lab are looking to recruit SHAFR members to participate in an IARPA-funded research project. We are developing AI systems designed to evaluate the risk of revealing national security secrets from aggregating large collections of less sensitive information. Whereas some argue this possibility may require radically reducing the availability of such information, our project aims to ensure government classification policies and procedures are based on rigorous peer-reviewed research. In this first stage of the project, we are organizing a workshop in which historians will evaluate whether and how AI can create a “mosaic” that reveals information that should be protected.

Participation will take place through a structured online workshop. It will include training in new AI-augmented research methods, such as discussion of prior work on the mosaic effect and demonstrations of research tools, like the FOIArchive search platform. No technical background is required, but participants need to have experience analyzing declassified documents, and some familiarity with U.S. foreign relations in the 1970s. Most of the time will be spent evaluating examples of AI-generated reconstructions of previously classified information from this period. Participants will be compensated and may have the opportunity to participate in future workshops.  

To apply, please complete this form, ideally by March 1, 2026, at 5pm ET. Should you have any questions, feel free to contact: [email protected]