The Pentagon is seeking $30.3 million over the next five years to develop an artificial intelligence-enabled successor to the government's traditional polygraph, according to a Department of Defense budget request. The program, known as Polygraph+ or Polygraph Next, would aim to improve how the U.S. military and security agencies assess truthfulness by combining scoring algorithms, machine learning and a technique described as "standoff sensing."
The proposal is notable not only for its modest price tag by defense standards, but for what it signals about the Pentagon's broader push to automate judgment in high-stakes settings. Polygraphs have long been controversial, widely used in government screening but persistently challenged by scientists, civil liberties advocates and many in the legal community. By adding AI to the mix, the Defense Department appears to be betting that data-driven models can extract more reliable signals from human behavior than conventional instruments that measure physiological responses such as heart rate, respiration and perspiration.
AI Meets Interrogation
The budget language points to a program focused on "scoring algorithms" that use artificial intelligence and machine learning to refine how examiners interpret results. In practice, that could mean software trained on large datasets of polygraph sessions, behavioral cues or other biometric inputs, with the goal of producing a more consistent assessment than a human examiner alone. The inclusion of standoff sensing suggests the Pentagon also wants to explore methods that do not require direct contact with the subject, potentially using remote sensors to detect subtle physiological or behavioral indicators.
That ambition places Polygraph+ at the intersection of frontier AI research and one of the most disputed applications of behavioral science. Supporters of automated screening argue that machine learning can identify patterns too complex or too faint for human observers to detect. Critics counter that AI systems are only as good as the data used to train them, and that in a domain as subjective and context-dependent as deception detection, automation may amplify error rather than reduce it.
Old Tool, New Risks
The polygraph has never been a definitive truth machine. Its results are generally considered inadmissible in many courts, and experts have long warned that stress, anxiety, medical conditions and examiner bias can distort outcomes. The Pentagon's interest in an improved version reflects a familiar government dilemma: the demand for faster, more scalable screening tools versus the scientific uncertainty surrounding whether deception can be measured reliably at all.
The move also arrives at a moment when AI is being pushed into more sensitive public-sector functions, from surveillance and intelligence analysis to hiring and fraud detection. Each use case raises a similar question: whether the technology is improving decision-making or merely giving it a veneer of objectivity. In the case of a lie detector, that concern is especially acute because the stakes can include security clearances, employment decisions and access to classified information.
The reference to standoff sensing is particularly significant. Remote sensing technologies can broaden the range of environments in which screening might occur, but they also intensify privacy concerns. If the Pentagon is exploring ways to infer truthfulness without direct physical contact, the program could eventually test the boundaries of what is considered acceptable observation in government screening.
What Comes Next
For now, Polygraph+ remains a budget request, not a fielded capability. The $30.3 million figure suggests a research and development effort rather than an imminent operational rollout. Still, the proposal is a clear indicator that the Defense Department sees value in pursuing AI-assisted deception detection despite decades of skepticism surrounding the underlying science.
The larger significance may lie less in whether the Pentagon can build a better lie detector than in what its effort reveals about the direction of defense technology. The department is increasingly willing to invest in systems that promise to convert ambiguous human signals into machine-readable scores. If Polygraph+ advances, it could become a test case for how far the U.S. government is prepared to trust algorithms in judgments that have traditionally depended on human interpretation — and how much uncertainty it is willing to accept in the name of security.