IC Technology & Cyber — 2026-10-07
Pentagon Retreats From AI Polygraph Trustworthiness Analysis After Scientific and Civil Liberties Pushback
BLUFDCSA's narrow denial sidesteps the language-based deception research, leaving Congress as the only near-term check on whether automated credibility tools survive scientific scrutiny.
In a Wednesday emailed statement, the Defense Counterintelligence and Security Agency said active research for its "Modernizing Polygraph" effort does not include generative AI, large language models,
Analysis
DCSA's disclaimer narrows the Modernizing Polygraph effort to physiological scoring and non-contact sensing, but it leaves the language-based deception models unaddressed. It does not say whether the Air Force and university text-analysis work has ended or moved to another office. Funded lines for AI scoring and standoff sensing remain, so scientific objections to automated credibility scoring still apply. DCSA may have drawn the line around "active" research to deflect scrutiny while deceptive-speech work continues under another name, sponsor or classification. Congress, which has not approved the budget, could require DCSA to define project scope before appropriating funds. Confidence is moderate: both sources trace to the same defense-trade reporting cluster, with no independent DCSA documentation.
DCSA's disclaimer narrows the Modernizing Polygraph effort to physiological scoring and non-contact sensing, but it leaves the language-based deception models unaddressed. It does not say whether the Air Force and university text-analysis work has ended or moved to another office. Funded lines for AI scoring and standoff sensing remain, so scientific objections to automated credibility scoring still apply. DCSA may have drawn the line around "active" research to deflect scrutiny while deceptive-speech work continues under another name, sponsor or classification. Congress, which has not approved the budget, could require DCSA to define project scope before appropriating funds. Confidence is moderate: both sources trace to the same defense-trade reporting cluster, with no independent DCSA documentation.
2 sources
- Has the Pentagon given up on AI polygraph analysis of trustworthiness? -
C4ISRNet - Pentagon AI lie detector -
MIT Technology Review