Pentagon seeks funding for AI lie detector program

A Department of Defense budget request proposes a new program called Polygraph+ or Polygraph Next to modernize federal lie detection tools. The effort would use AI and machine learning scoring algorithms alongside “standoff sensing,” a technique intended to collect physiological readings without attaching devices to a subject.

The Defense Counterintelligence and Security Agency would run the program, which is intended for vetting prospective employees and detecting insider threats. The plan comes as the Pentagon has increased polygraph use during leak investigations under Defense Secretary Pete Hegseth, including tests of around 50 Joint Staff officers after reporting on depleted US weapons stockpiles in the war with Iran.

Potential technology directions can be seen in earlier Pentagon work. The Defense Innovation Unit selected Presage Technologies and Altec Research in 2023 to build deception detection prototypes, including systems that measure heart rate, breathing, head movement, facial skin temperature, and pore activity using cameras or non-contact sensors.

Researchers remain skeptical, noting that polygraph results have long faced challenges over accuracy, subjectivity, bias, and countermeasures. Congress’s Office of Technology Assessment found in 1983 that evidence for employee screening was very limited, and the US National Research Council said in 2003 that efficacy evidence was “weak at best.” Critics argue AI cannot solve the core problem: there is no reliable physiological marker that proves someone is lying.