Pentagon Budgets $6.4M for AI Polygraph; Contractor TBD

The Pentagon’s FY2027 budget justification requests $6.421 million for a program called Polygraph Next, which it says will integrate AI and machine-learning scoring with non-contact sensing, and its five years of requests add up to about $30.3 million by our own sum. Those are the agency’s own budget figures from an April 2026 document, and we checked them only against its tables. The contractor and contract method are listed as TBD.

What the Budget Document Says

The Defense Counterintelligence and Security Agency’s FY2027 RDT&E justification, dated April 2026, lists a project called Polygraph Next under program element 0305146V, Defense Joint Counterintelligence Activities, at R-1 line 269.

It describes Polygraph+/Polygraph Next as “a transformative effort to modernize federal polygraph and credibility assessment technologies,” in which the agency “will conduct independent validation studies, field testing, and prototype evaluations to advance standoff sensing, automated scoring, and decision aids.”

It says that “by integrating non-contact sensing systems and AI/ML-based scoring algorithms” the program will address personnel vetting and insider threat detection. It names research partners ARLIS and Oak Ridge National Laboratory “and industry leaders.”

The Money

The document lists no funding for Polygraph Next in prior years, FY2025 or FY2026. It requests $6.421 million for FY2027, then $6.449 million, $6.158 million, $5.660 million and $5.654 million for FY2028 through FY2031, with cost to complete and total cost both marked “Continuing.”

This publication’s own arithmetic: the five requested years sum to $30.342 million. MIT Technology Review reported the program as $30.3 million over five years from this request, which matches.

The cost-analysis table lists a single line, with contract method and type given as “C/TBD” and the performing activity as “TBD.” It gives no award date.

What FY2027 Pays For

For FY2027 the document says the program plans “to scale production, necessary to conduct operational pilots with interagency polygraph programs to determine performance in the various operational applications while utilizing a broad spectrum of polygraph test formats.”

It says the program will also “begin developing cloud-based storage and a suite of tools to conduct data analytics to refine software, address gaps, and to conduct research necessary for identifying new diagnostic features.” Its increase statement says the rise from FY2026 enables development of future capabilities and expanded cloud storage and analytics.

The schedule-details page lists a Polygraph Next entry running from quarter 1 of 2028 to quarter 3 of 2029, without naming an event.

What the Document Does Not Say

The project pages this publication read give no accuracy or reliability figure for the technology they propose, name no vendor, and do not say how the AI and machine-learning scoring would be validated beyond the independent validation studies they list.

It is a budget estimate. The project pages we read do not say whether Congress has funded the line.

The Structural Read

The document says the agency will conduct independent validation studies, field testing and prototype evaluations to advance standoff sensing, automated scoring and decision aids. It lists no funding for Polygraph Next in prior years, FY2025 or FY2026, then $6.421 million in FY2027 and roughly $5.7 to $6.4 million a year through FY2031, with cost to complete marked Continuing.

For FY2027 the document says the program plans to scale production, which it says is necessary to conduct operational pilots with interagency polygraph programs, and to begin developing cloud-based storage and a suite of tools to conduct data analytics. Its cost-analysis table lists a single line, with contract method and type given as C/TBD and the performing activity as TBD, and gives no award date.

The pages this publication read name no vendor and give no accuracy or reliability figure for the technology. They name ARLIS and Oak Ridge National Laboratory as research partners and refer to industry leaders. The project pages we read do not say whether Congress has funded the line.

Three Implications

WHAT KIND OF DOCUMENT IT IS A budget estimate. The figures are the agency’s own request in a document dated April 2026, and the project pages we read do not say whether Congress has funded the line.

THE CONTRACT FIELDS Contract method and type are listed as C/TBD and the performing activity as TBD, with no award date. The project pages we read name ARLIS and Oak Ridge National Laboratory as research partners and refer to industry leaders, but name no vendor.

WHAT THE PAGES LEAVE OUT The project pages give no accuracy or reliability figure for the proposed technology and do not say how the AI and machine-learning scoring would be validated beyond the independent validation studies they list. Nothing here says whether the technology works or how it would be used on any person.

What Is Not Established

This publication read the Defense Counterintelligence and Security Agency’s budget justification and relied on MIT Technology Review only for the five-year figure, which it checked against the document’s own table. It did not seek a response from the agency.

Nothing here says whether the technology works or how it would be used on any person. All figures are the agency’s own request, and nothing in this piece is investment advice.

Business Engineer Framework

The Map of AI — The Stack Beneath the Applications

The Map of AI places more than 200 companies across nine layers of the stack, from silicon to application. It shows the layers that sit beneath the applications people use every day.