The US government plans to invest $30.3 million over five years in an upgraded lie detection system, according to a Department of Defense budget request. The program, dubbed Polygraph+ or Polygraph Next, will center on scoring algorithms powered by artificial intelligence and machine learning, alongside a method called "standoff sensing" that can capture physiological readings without physically attaching devices to subjects. The budget document says the project aims to "modernize federal polygraph and credibility assessment technologies" to boost their accuracy and reliability, though experts warn it may represent just another failed attempt to use technology for detecting deception.
The initiative arrives during a period of heightened tension inside the department, as Defense Secretary Pete Hegseth has ramped up polygraph testing to locate sources of suspected press leaks. In September, the New York Times reported that roughly 50 Joint Staff officers underwent polygraph examinations following news stories about depleted US weapons stockpiles in the conflict with Iran. The Defense Counterintelligence and Security Agency, which handles federal background checks, will oversee Polygraph+ and deploy the technology for prospective employee screening and "insider threat detection," according to the budget document awaiting congressional approval. Earlier Pentagon efforts offer hints about potential technologies: in 2023, the Defense Innovation Unit selected Presage Technologies, which claims to measure heart rate and breathing through standard cameras, and Altec Research, a medical sensor firm developing non-contact sensing tools. A screenshot of Altec's prototype shows it monitors head movement, facial skin temperature, and pore activity.
Traditional lie detection technology has remained largely unchanged since the polygraph's invention in the 1920s, with examiners relying on blood pressure, pulse, breathing, and perspiration measurements to judge truthfulness. They assess whether respondents are lying based on differences between physiological responses to baseline questions like "Is the sky blue?" and target questions such as "Have you ever committed a crime?" The federal government administers tens of thousands of these tests annually for employee screening, but the technology's reliability has faced repeated challenges and its results are seldom admissible in court. In 1983, Congress's Office of Technology Assessment found very limited evidence supporting polygraph use for employee screening, while in 2003, the US National Research Council declared the evidence for its effectiveness "weak at best." Research indicates humans can detect lies just over half the time without technical help, and though the American Polygraph Association claims 80% to 94% accuracy, the 2003 NRC report noted that a screening test at this accuracy level could still generate many errors—potentially falsely accusing tens of thousands among the DOD's 2.8 million employees. Additional problems include subjective interpretations that vary widely between examiners, higher rates of minority groups being judged deceptive, and the ability of trained subjects to employ countermeasures like stepping on a hidden pin to artificially spike responses to baseline questions.
AI could theoretically enhance polygraphs by identifying patterns human examiners miss, and algorithms might enable "multi-modal" deception detection that combines multiple measurements into a single score harder to manipulate. "The more you can have combined methods that approach it from these three different angles, the more successful you will be," says Sophie van der Zee, an associate professor studying deception at Erasmus University in Rotterdam, referring to physiological stress, cognitive load, and conscious concealment efforts. Yet various new lie detection strategies attempted over decades—from thermal cameras to pupil trackers to brain scans—have failed to produce reliable results outside laboratory settings. The fundamental obstacle remains: there's no universal indicator of lying that applies to everyone consistently. "There is still no Pinocchio's nose," van der Zee notes. Past projects like Silent Talker and AVATAR, which incorporated eye tracking, voice analysis, and body movement detection, have quietly disappeared. Legal scholars warn that combining AI with polygraphs creates "the worst of both worlds" because machine learning cannot reliably connect patterns to lying without genuine ground truth—even with vast polygraph records, there's no way to verify whether those tests were accurate.
Marion Oswald, a law professor who has written on polygraph use in justice systems, worries new lie detection forms will function more as psychological tools than scientific instruments. "It seems very much a response to the concern of the current administration to leaks and perceived lack of loyalty," Oswald says, adding that lie detection is "being used as a threat, to intimidate and force people to confess to things, as opposed to anything that's actually getting valid information." Van der Zee points out the machine's primary impact is deterrence, with subjects often confessing before testing begins, but cautions "that only works if people think a polygraph works." The effectiveness of intimidation-based approaches depends entirely on sustained belief in the technology's validity, a foundation that decades of scientific scrutiny have steadily eroded. Organizations investing heavily in unproven detection systems may find themselves managing perceptions rather than uncovering truth, while the human cost of false accusations at scale could undermine the very trust and loyalty these programs ostensibly seek to protect.