Pentagon Requests $30.3 Million for AI-Based Lie Detection Technology
The Pentagon has announced a request for $30.3 million over the next five years to develop an innovative AI lie detector system called Polygraph+ or Polygraph Next. This initiative is included in the Defense Counterintelligence and Security Agency's (DCSA) budget proposal for 2027, which aims to modernize federal polygraph and credibility assessment technologies.
The DCSA's budget outlines a plan for standoff sensing, which involves taking physical readings from individuals without the need for direct contact. This technology will leverage artificial intelligence and machine learning to enhance scoring and decision-making processes. The funding request includes $6.42 million for 2027, followed by annual allocations ranging from $5.65 million to $6.45 million through 2031, pending congressional approval.
The primary goal of this program is to improve the vetting of personnel and bolster insider threat detection capabilities. The DCSA also plans to incorporate cloud storage and data analysis tools to refine the software's effectiveness. Under the leadership of Defense Secretary Pete Hegseth, the Pentagon has increasingly relied on polygraph tests to identify sources of information leaks, with reports indicating that approximately 50 officers on the Joint Staff underwent testing in September.
The Pentagon has previously explored similar technologies. In 2023, its Defense Innovation Unit selected two companies to create prototypes for this type of AI-driven lie detection. Presage Technologies claims its system can measure heart and breathing rates using standard cameras, while Altec Research's prototype monitors head movements, facial skin temperature, and pore activity.
It is important to note that traditional polygraphs do not directly detect lies; rather, they measure physiological responses such as breathing, pulse, blood pressure, and sweat, which are then interpreted by an examiner. A 2003 report from the US National Research Council described the evidence supporting the accuracy of polygraphs as "weak at best."
Accuracy and Skepticism Surrounding Polygraphs
The American Polygraph Association asserts that polygraph tests can achieve an accuracy rate between 80% and 94%. However, this level of accuracy still raises concerns about potential errors, especially given the Pentagon's workforce of approximately 2.8 million individuals.
Experts have expressed skepticism regarding the integration of AI into polygraph technology. Kyri Kotsoglou, a legal scholar at Northumbria University, argues that adding AI to the polygraph could exacerbate existing uncertainties. Marion Oswald, a law professor, emphasizes that even with extensive records from polygraph tests, the validity of those tests remains questionable.
On the other hand, Sophie van der Zee, a deception researcher at Erasmus University, suggests that AI could potentially consolidate various measures into a single score, making it more challenging to manipulate.
Clearance Granted
Historically, previous AI lie detection initiatives have not been successful. For instance, the EU-funded iBorderCtrl project, which aimed to assess deception through video analysis, ultimately failed, similar to the US AVATAR border project.
Currently, various AI technologies are being utilized to analyze human behavior, including AI voice analysis in US prisons, facial recognition systems in UK retail environments, and AI-driven camera surveillance by US law enforcement agencies.
