Pentagon Requests $30.3 Million for AI-Powered Lie Detection Program

The Pentagon has submitted a budget request of $30.3 million over five years for the Polygraph+ program, which aims to improve lie detection technologies. This initiative is overseen by the Department of Defense (DOD) and run by the Defense Counterintelligence and Security Agency (DCSA). The funding will support prototype development by Presage Technologies and Altec Research, both selected for their expertise in deception detection.

Historically, polygraph tests have been controversial, with the American Polygraph Association claiming an accuracy rate between 80% and 94%. However, the National Research Council (NRC) reported in 2003 that the evidence supporting polygraph efficacy was "weak at best," while a 1983 report from Congress’s Office of Technology Assessment concluded that there was very limited evidence for the use of polygraphs in screening.

The DOD employs approximately 2.8 million people and conducts tens of thousands of polygraph tests annually for employee screening. Despite the claims of accuracy, experts like Kyri Kotsoglou from Northumbria Law School argue that the effort to simplify lie detection into a quantifiable measure is misguided. Marion Oswald, a law professor, emphasizes that even comprehensive records from polygraph tests do not guarantee their reliability.

This funding request follows the Defense Innovation Unit's (DIU) open submission process earlier in 2023 for deception detection technologies, indicating a renewed interest in enhancing the capabilities of lie detection systems. The Polygraph+ program is positioned within a broader context of evolving technologies aimed at improving security and screening processes, including initiatives like AVATAR for border detection and the EU-funded iBorderCtrl pilot program.

As the Pentagon moves forward with this budget request, the implications for practitioners in security and intelligence may include access to more advanced tools for deception detection, though the efficacy of such technologies remains debated.