The Pentagon wants $30 million to build an AI-powered lie detector
MIT Technology Review Amit Katwala
The Pentagon wants $30.3 million for an AI-powered lie detector called Polygraph+. It’s meant to modernize a test that’s been shaky for decades and is still easy to beat.
Based on reporting by MIT Technology Review, Amit Katwala — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
The Pentagon wants to spend $30.3 million over the next five years on a new attempt to make lie detection less flimsy. The program has two names in the budget request: Polygraph+ and Polygraph Next. Its pitch is simple enough. Use artificial intelligence, machine learning, and “standoff sensing” to improve the old polygraph’s accuracy and reliability without wiring people up as much as before.
The work would sit with the Defense Counterintelligence and Security Agency, which handles federal background checks. According to the budget document, the technology would be used for vetting prospective employees and for “insider threat detection.” The request has not been approved by Congress yet, and the agency has not said which specific tools it plans to use. But the Pentagon has already been testing the waters.
In 2023, the Defense Innovation Unit asked companies for deception-detection products and picked two to build prototypes. One was Presage Technologies, which says it can measure heart rate and breathing with ordinary cameras. The other was Altec Research, a medical sensor company moving into non-contact sensing. A screenshot released by the DIU showed Altec’s prototype tracking head movement, facial skin temperature, and pore activity. None of those groups responded to requests for comment, and the DIU declined to comment.
The push is happening against a messy backdrop. Under Defense Secretary Pete Hegseth, the Pentagon has leaned harder on polygraph tests while hunting for leaks. In September, the New York Times reported that around 50 officers on the Joint Staff were tested after coverage of depleted US weapons stockpiles in the war with Iran. That is the sort of environment where a new lie detector looks attractive, even if the old one has been questioned for years.
And it has been questioned for years. The modern polygraph still rests on blood pressure, pulse, breathing, and sweat, usually compared between baseline and target questions. It is widely considered unreliable, rarely admissible in court, and has been repeatedly criticized by official reviews. People can also learn countermeasures to game it. AI might find patterns humans miss, but the basic problem remains: there is no clean way to prove that a physiological signal means someone is lying.
My take — AI-written commentary, not fact-checked reporting
This is classic Pentagon thinking: if the old gizmo is shaky, add more sensors and call it modernization. But layering AI on top of a tool that can’t prove what it claims only gives bureaucracy a shinier prop. The real use here is pressure, not truth, and everyone in the room knows it.
Read more about this at: MIT Technology Review
