'Synthetic suspicion': Laperruque maps AI-prompted police stops
TL;DR
- Laperruque documents stops and raids triggered by AI flags from Border Patrol license-plate scans, DHS financial-pattern targeting, and ICE's Palantir-built ELITE system.
- Rep. Bennie Thompson says ICE treats Mobile Fortify biometric matches as 'definitive,' authorizing officers to ignore a US birth certificate.
- The essay calls for pausing AI use in police decision-making until independent vetting, procurement standards, and training safeguards are validated.
A ranking member of the House Homeland Security Committee says ICE's own policy lets an officer disregard a US birth certificate if the Mobile Fortify facial-recognition app pings the person as a non-citizen.
That claim anchors Jake Laperruque's essay at Tech Policy Press. Laperruque, a fellow at the outlet, argues that American policing is drifting into what he calls 'synthetic suspicion': the use of AI outputs to generate the justification for stops, searches, and arrests.
Rep. Bennie Thompson, as Laperruque quotes him: 'ICE officials have told us that an apparent biometric match by Mobile Fortify is a "definitive" determination of a person's status and that an ICE officer may ignore evidence of American citizenship… if the app says the person is an alien.'
He walks the pattern across agencies. AP's Byron Tau and Garance Burke documented a Border Patrol program that scans license-plate reader data and flags 'suspicious' driving patterns; the ACLU's Jay Stanley collected the account of a woman pulled over on a multi-state road trip in September 2025 because an AI had flagged her, with further stops in California, Michigan, Texas, and Wisconsin. 404 Media's Joseph Cox has reported on DHS 'Predictive Intelligence Targeting Teams' that pick subjects by 'financial activity patterns.'
Palantir's ELITE system, used by ICE to build deportation target dossiers, drew a flat description from a deportation officer in an Oregon court filing. 'It tells you how many people are living in this area and what's the likelihood of them actually being there… It's kind of like Google Maps,' the officer said, acknowledging he did not know what data was driving the recommendations.
Alongside those stops and raids sit three wrongful arrests from facial-recognition mismatches over the past year, each involving months-long detentions: a North Dakota case covered by the New York Times in March 2026, a Jacksonville case in which a ten-year-old's father was taken in, and a Maryland case documented by ACLU-MD. A 2019 Washington County, Oregon reconfiguration had set the system to return five candidates every time regardless of match quality. A 2018 NYPD detective, in a robbery case, skipped the lineup and texted a witness a facial-recognition hit with the caption, 'Is this the guy?' A Chicago Police training document, Laperruque notes, warned officers: 'Do not let the software become exculpatory evidence.'
His recommendation is a pause. No AI in police decision-making, Laperruque writes, until independent vetting, procurement standards, and training safeguards are in place. Two analysts we track were circulating the piece within hours of publication.
Shared on Bluesky by 2 AI experts
-
Tech Policy Press fellow Jake Laperruque examines “synthetic suspicion” — when an AI system generates the justification for police stops, searches and arrests — and argues policymakers should act before it becomes a norm…
View on Bluesky →
Originally reported by techpolicy.press
Read the original article →Original headline: “Synthetic Suspicion” and the Perils of AI-Prompted Police Actions