What happened

The U.S. Department of Defense, often called the Pentagon, is seeking money to update federal lie-testing tools. Its proposed program is called Polygraph Next, also described as Polygraph+. A fiscal-year 2027 budget request allocates $6.421 million for the first year. The listed amounts for 2027 through 2031 add up to about $30.3 million.

That does not mean a finished AI lie detector is ready. It is a research and development plan. The program would test new sensing methods, automated scoring, and software that helps officials assess people.

Background

A polygraph records bodily responses while a person answers questions. The result is used to judge how credible an answer seems. Polygraph Next would add standoff sensing, meaning measurements that do not require sensors to touch the person directly. It would also use AI and machine-learning scoring algorithms.

The budget document says the program aims to improve accuracy, reliability, usability, and traceability. It also calls for centralized data storage and analytics. In plain terms, officials want to collect information in a more consistent way, turn it into scores, and keep records that can be reviewed later.

Why it matters

The proposed use is not a classroom experiment. It is tied to personnel vetting, or checking people who may receive access to sensitive work. The document also mentions insider-threat detection. That means looking for risks that could come from people inside an organization.

An automated score could make screening faster or easier to compare. But a score is not the same as proof. If a system labels someone as deceptive, the consequences could affect a job or access to sensitive information. The most important question is therefore not whether AI sounds clever. It is how the system is tested, explained, challenged, and overseen.

The plan itself recognizes this problem. It includes independent validation studies, field testing, and prototype evaluations. Those steps show that the government is proposing a capability to test, not announcing a proven way to read the truth.

What is confirmed

The public budget papers describe pilots with other government polygraph programs. They say the program would use several polygraph test formats. For fiscal year 2027, officials also plan cloud-based storage and data-analysis tools. Those tools would help refine the software, find gaps, and study possible diagnostic features.

The papers name the Applied Research Laboratory for Intelligence and Security, Oak Ridge National Laboratory, and industry partners as research collaborators. They do not name a final contractor. The request is also a plan for funding, not proof that every proposed activity will happen. The details are available in the public budget justification.

What remains unknown

The documents do not provide a final accuracy result or a clear performance target. They do not explain which sensors will be used, which signals will matter most, or how the system will handle stress, illness, or unusual reactions. They also do not say how long data would be kept, who could inspect it, or whether a human could overturn an AI score.

Most importantly, the budget request does not establish that the technology can reliably detect lies. That question is supposed to be answered through testing.

What to watch next

The next milestones are budget approval, a contract, and the design of the independent studies. The results should show where the tool works, where it fails, and how often it gives the wrong impression. Watch also for rules about data access and human review. Until then, Polygraph Next is best understood as a government research program with a high-stakes goal, not a finished truth machine.