Texas’ expanding AI surveillance state outpaces legislative oversight

In recent years, the Texas Department of Public Safety (DPS) has quietly constructed a powerful and far-reaching surveillance network that increasingly relies on AI to monitor people, places, and activities across the state.

Powered by Governor Greg Abbott’s $11 billion Operation Lone Star, a sweeping border enforcement program, DPS has acquired a range of tools that now form one of the most technologically advanced domestic intelligence infrastructures in the nation. The Texas Standard said these developments have taken place largely under the radar with minimal legislative oversight and only the faintest nod to civil liberties.

The arsenal includes AI software using facial recognition that searches through billions of scraped images and multimillion-dollar contracts with private vendors that grant the agency access to databases, surveillance platforms, and forensic tools. Records obtained by the Texas Observer show that DPS has used emergency declarations and executive orders to acquire tools that would be controversial even in military or intelligence settings, including technologies originally built for combat zones and national security threats.

Among the most prominent tools is Tangles, an AI-powered platform that mines publicly available digital spaces and commercial data sources. Since 2021, DPS has spent nearly $1 million on Tangles licenses, and has used the software to identify what the agency calls potential domestic terror threats or mass casualty events. The tool’s capabilities include cell phone location tracking without a court order, a feature that legal experts argue could violate the Fourth Amendment. Civil liberties attorneys say the software, used in concert with other AI tools, gives law enforcement sweeping, warrantless insight into individuals’ lives – where they go, who they speak with, and what they post online. The full extent of how Tangles is deployed remains unknown, as DPS has withheld key records and declined to answer specific questions about the software’s application.

Facial recognition has emerged as another critical component of DPS’ surveillance efforts. Through contracts with Clearview AI the agency has access to a system trained on over 40 billion images scraped from across the Internet. Though Texas law prohibits biometric data mining without explicit consent, the state’s own use of Clearview under Operation Lone Star has not only continued but expanded, with a $1.2 million contract extending use through 2030.

Google, currently being sued by Texas for biometric privacy violations, has used the state’s Clearview contract to argue that Texas is selectively enforcing its privacy laws. Clearview maintains that its technology is used solely for criminal investigations and not in active surveillance, but critics note that the lines between those categories are often deliberately blurred.

DPS’ surveillance matrix doesn’t end there. It also uses Cellebrite, a digital forensics tool capable of cracking encrypted phones and extracting massive quantities of data. Cellebrite’s AI can analyze photos and data from phones to assist in investigations, and DPS has renewed its license through 2027 for $2.7 million.

Another prong of the state’s approach is automated vehicle tracking, including access to Motorola’s LEARN database and Flock Safety’s growing license plate camera network. These tools allow state police to track vehicle movement in real time and generate “vehicle fingerprints” based on make, model, and other characteristics. With more than 3,800 cameras in Houston alone, Flock’s network has become deeply embedded in Texas law enforcement’s intelligence capabilities.

Adding to its reach, DPS manages Operation Drawbridge, a sprawling network of over 9,000 wildlife cameras across South Texas that produces more than 250,000 images per day. In the past, humans were required to review this footage; now, the task has been handed over to AI. In 2023, DPS entered a $6 million contract with Deloitte Consulting to develop an AI system that classifies and flags images based on whether they include people, weapons, drugs, or other elements of interest. The agency’s internal inventory suggests that the AI may eventually be used for predictive analysis, attempting to forecast future movements or threats.

Supporting this interconnected network is Spart-N, DPS’ proprietary analytics platform that integrates data streams from all of these systems. DPS also receives access to Verus, software that monitors and transcribes inmate phone calls, as well as ShadowDragon’s SocialNet and Dataminr’s FirstAlert, both of which analyze data from platforms like X. All told, DPS is expected to spend more than $20 million over the next five years on social media monitoring tools alone.

Despite the enormous scale and reach of DPS’ surveillance efforts, the Texas Legislature has done little to rein in its growth. Lawmakers formed an AI advisory council last session to gather information on how agencies are using artificial intelligence, but the council lacks enforcement authority and proposed laws have largely failed to impose binding constraints.

Representative Giovanni Capriglione initially filed a bill, the Texas Responsible AI Governance Act, which would have prohibited certain uses of facial recognition and required transparency from AI vendors, but it was replaced with a significantly weakened version, House Bill 149, which strips out many of its original regulatory provisions and explicitly bars the AI advisory council from interfering with agency operations.

Language in the revised bill suggests that AI systems cannot be used to uniquely identify individuals if doing so infringes on constitutional or legal rights, but it leaves open to interpretation what constitutes such an infringement. Experts say this ambiguity effectively allows law enforcement to continue using tools like Clearview without restrictions so long as they assert compliance with constitutional protections. No explicit ban exists, and no provisions enable individuals harmed by AI misuse to take legal action against state agencies or technology vendors.

Senator Tan Parker introduced a complementary bill, SB 1964, which would require agencies to conduct impact assessments when deploying AI, but even these reports would be shielded from public disclosure by being exempted from the Texas Public Information Act. Critics argue that without transparency or enforcement mechanisms, the legislation amounts to little more than paperwork.

Public concern about AI surveillance in Texas is rising across the political spectrum. Organizations as ideologically distant as the ACLU of Texas and the Texas Public Policy Foundation are finding common ground in opposing warrantless surveillance. Both groups have warned that the unchecked growth of police technology undermines constitutional protections and risks establishing the foundations of a surveillance state.

Republican lawmakers like Brian Harrison have voiced alarm at the growing partnership between state government and private tech vendors. After reports surfaced about warrantless phone tracking by DPS, Harrison pledged to investigate and later introduced a bill requiring a warrant for license plate reader usage and similar tracking technologies.

Capriglione has expressed deep concerns over the sweeping nature of these tools. At a recent policy forum he warned that Texas is on a path toward a surveillance society marked by license plate readers on every street and databases that log the movements of people who have committed no crime. His statements reflect a growing awareness that technology, while useful for crime prevention, poses immense dangers when used without meaningful oversight.

Yet DPS continues to resist accountability. The agency has refused to release memos regarding its use of Clearview AI and insists that internal systems include audit logs, though it won’t share them. DPS leaders have claimed in testimony that the technologies are used only to assist with investigations and are not deployed for proactive mass surveillance. But public records suggest otherwise, and the reluctance to disclose even basic operational guidelines only deepens concerns.

Legal experts warn that the normalization of surveillance tools threatens to reshape American law enforcement in fundamental ways. These systems may not carry guns or wear badges, but they can quietly identify, track, and profile individuals in ways that were unimaginable a decade ago. And as lawmakers debate cosmetic reforms while approving massive new contracts, the infrastructure for a full-fledged surveillance state continues to grow.

Article Topics

biometrics  |  facial recognition  |  Texas  |  United States  |  video surveillance