Pentagon Seeks $30.3 Million for AI-Powered Polygraph+

The U.S. government is seeking $30.3 million over five years for Polygraph+, also called Polygraph Next, through a Department of Defense budget request that Congress has not yet approved. [1] The program would focus on AI and machine-learning scoring algorithms plus “standoff sensing,” which takes physiological readings without attaching a device to the subject. [1] The budget document says the project would modernize federal polygraph and credibility-assessment technologies to improve their accuracy and reliability. [1]

The DCSA would run Polygraph+ for prospective-employee vetting and “insider threat detection,” but the specific technologies have not been disclosed. [1] Under Defense Secretary Pete Hegseth, the Pentagon has increasingly turned to polygraph tests while trying to identify sources of alleged press leaks. [1] Multiple scientific bodies, including the National Research Council, have found polygraph evidence weak or unreliable, and experts say AI compounds rather than fixes the lack of universal physiological markers for deception. [2]

Polygraph+ combines two bets: that algorithmic scoring can improve credibility assessment and that contactless sensing can collect useful physiological signals. The central test is not simply whether AI can automate scoring, but whether the measurements reliably distinguish deception; the scientific criticism cited by the sources says that foundation remains unresolved. With congressional approval pending and the technical stack undisclosed, scrutiny can focus on validation standards, error rates, and how results would affect employee vetting. What the sources don't address: which sensors, models, training data, independent tests, or appeal procedures the DCSA would use.