The Pentagon wants $30 million to build an AI-powered lie detector
2026-09-28 · MIT Technology Review
The Pentagon wants $30 million to build an AI-powered lie detector
The US government plans to spend $30.3 million over the next five years on an improved lie detector, according to a Department of Defense (DOD) budget request. The program, named Polygraph+ or Polygraph Next, will focus on scoring algorithms using artificial intelligence and machine learning, alongside a technique called "standoff sensing," which takes physiological readings without attaching a device to the subject.
Background and Objectives
According to budget document details first reported by Inside Defense, the project aims to "modernize federal polygraph and credibility assessment technologies" to improve accuracy and reliability. The budget has not yet been approved by Congress.
The program will be run by the Defense Counterintelligence and Security Agency (DCSA), which conducts federal background checks. The new technology will be used for vetting prospective employees and "insider threat detection." Specific technologies remain unclear, and DCSA did not respond to requests for comment.
This move comes amid high tension within the department. Under Defense Secretary Pete Hegseth, the Pentagon has increasingly used polygraph tests to find alleged press leakers. In September, the New York Times reported that around 50 Joint Staff officers were given polygraph tests following news coverage on the depletion of US weapons stockpiles.
Potential Technological Clues
Other Pentagon efforts offer clues. In 2023, the Defense Innovation Unit (DIU) sought companies with deception detection products. It selected two to build prototypes:
- Presage Technologies: Claims to measure heart and breathing rates using standard cameras.
- Altec Research: A medical sensor company branching into non-contact sensing. A DIU screenshot shows Altec’s prototype tracks head movement, facial skin temperature, and pore activity.
Limitations of Traditional Polygraphs
Current lie detection technology has barely changed since the 1920s. Examiners rely on blood pressure, pulse, breathing, and sweat measurements, judging veracity by comparing physiological responses to baseline questions ("Is the sky blue?") and target questions ("Have you ever committed a crime?").
While the federal government conducts tens of thousands of tests yearly, the technology's reliability is repeatedly challenged, and results are rarely admissible in court:
- Accuracy Concerns: In 1983, Congress’s Office of Technology Assessment found very limited evidence supporting polygraph screening. In 2003, the US National Research Council (NRC) called evidence for its efficacy "weak at best."
- Scale Risks: The DOD employs 2.8 million people; an imperfect system at that scale could falsely accuse tens of thousands.
- Subjectivity and Bias: Interpretations are often subjective, with different examiners getting wildly different results. Minority groups are more likely to be judged deceptive.
- Countermeasures: Interviewees can learn countermeasures to beat the test, such as artificially heightening physiological responses to baseline questions by stepping on a hidden pin.
"If you know how it works, you can beat it," says Sophie van der Zee, an associate professor at Erasmus University. She notes the machine's biggest effect is deterrence—subjects often confess beforehand—but that only works if people believe it functions properly.
Can AI Bring a Breakthrough?
Various new lie detection strategies have been attempted over decades, from thermal cameras to brain scans, but none yielded reliable results outside the lab. The problem is the lack of a single telltale sign of lying true for everyone. "There is still no Pinocchio’s nose," says van der Zee.
AI could theoretically improve polygraphs by finding data patterns examiners miss. AI algorithms are likely to be used for "multi-modal" deception detection, combining multiple measurements into an overall deception "score" harder to game. Van der Zee explains that lie detection targets three underlying factors: physiological stress, cognitive load, and conscious efforts to conceal lying. Current technology tackles only one.
"The more you can have combined methods that approach it from these three different angles, the more successful you will be," she says.