US Defense Security Cooperation Agency advances $31M polygraph project
- The United States Defense Security Cooperation Agency is advancing a $31 million project known as Polygraph Modernization to update military security screening procedures, according to a report.
- Current defense budget records show that $31 million is allocated for the Polygraph+/Polygraph Next project.
- Air Force has collaborated with the Defense Security Cooperation Agency and research institutions on deep learning models since 2019.
The United States Defense Security Cooperation Agency is advancing a $31 million project known as Polygraph Modernization to update military security screening procedures, according to a report. Current research documents spanning from 2021 to 2025 indicate that artificial intelligence will play a role in supporting the analysis of deceptive answers during military vetting.
Defense Budget and Technical Profile
Current defense budget records show that $31 million is allocated for the Polygraph+/Polygraph Next project. The system integrates automated scoring algorithms, decision-support tools, and specialized thermal cameras designed to monitor blood flow changes when a test subject experiences psychological stress.
Documents obtained by Defense News reveal that the U.S. Air Force has collaborated with the Defense Security Cooperation Agency and research institutions on deep learning models since 2019.
During a mock trial conducted in May 2025, a test subject sat before a virtual investigator while camera apertures and acoustic sensors recorded visual and audio data. This information was translated into large language models for sentiment analysis, while a human investigator monitored the situation from an external room and entered questions into a control system.
Assessing the Reliability of Machine Learning Algorithms
National security attorney Mark Zaid pointed out that traditional polygraphs record basic physiological responses such as respiration rate, sweat conductivity, and blood pressure, leaving evaluation authority entirely to human specialists. The key distinction in the new project is that artificial intelligence algorithms perform direct evaluations and classify subjects.
However, independent scientific data has not yet established superior reliability for machine learning-based deception detection.
Implications for Military Personnel and Courts
Inaccurate conclusions from automated testing carry direct consequences for military personnel. An officer who fails the examination faces a suspension of classified information access for at least one year, along with potential career impacts such as frozen promotions or exclusion from overseas combat deployments.
Scheffer that excluding polygraph results from military courts is entirely constitutional because the technology lacks scientific consensus.
Joint Staff underwent polygraph examinations as part of an investigation into information leaks concerning missile and interceptor stockpiles. Amid questions regarding accuracy, a Department of Defense official stated that results generated by automated algorithms are currently classified as supplementary information intended for specialist reference.
