Clearview AI contract links Army special forces to wider intelligence ecosystem

A small U.S. Army special forces purchase of Clearview AI facial recognition licenses has exposed a broader defense intelligence pipeline connecting biometric search, commercial data, AI-enabled entity resolution, former U.S. intelligence personnel, and a company that is building a separate Pentagon platform for operational data fusion.

Biometric Update reported in February on the underlying Army special forces Clearview solicitation and renewal. Independent journalist Jack Poulson later reported that Octaris Technologies served as the intermediary for the March purchase, identifying the Virginia Beach, Virginia company as the vendor on the award.

Separately, Biometric Update reported in January that USSOCOM was laying the groundwork for a broader expansion of battlefield identity intelligence and exploitation capabilities.

The USSOCOM request for information sought industry input on facial recognition, voice identification, tactical site exploitation tools, rapid DNA analysis, and AI and machine-learning analytics, underscoring that the Clearview purchase sits within a larger special operations push toward identity-driven intelligence.

Procurement records [here and here] identify Octaris as the vendor on a March award for Clearview AI service for the 1st Special Forces Command (Airborne), the Fort Bragg-based Army special operations headquarters that includes the Army’s various special forces groups.

The 1st Special Forces Command is subordinate to the U.S. Army Special Operations Command, the largest component of USSOCOM.

The award summary lists a total potential contract value of $339,415. The summary identified $78,750 as the March 16 purchase order amount for five Clearview AI licenses, which appears to correspond to the base-year obligation rather than a separate transaction.

The base period runs from March 20, 2026, to March 19, 2027, with three one-year option periods that could extend the subscription through March 19, 2030.

A February memorandum for the record describes the requirement as a follow-on action, funded with FY 2026 Regular Army Operation and Maintenance money and structured as a fixed-price brand-name buy.

The justification is unusually direct about why the command wanted Clearview and not a competing facial recognition tool.

“Without Clearview AI, DoD cannot rapidly analyze vast amounts of facial data, which makes it challenging to identify high-value targets with the required rate of accuracy needed for 1st Special Forces Group,” the Army memorandum states.

The documents identify the requiring activity as the 1st Special Forces Command (Airborne), but the justification document refers to the 1st Special Forces Group.

The distinction matters.

1st Special Forces Command is the Fort Bragg-based headquarters that includes Special Forces, civil affairs, psychological operations, and support units. The 1st Special Forces Group is an Indo-Pacific-focused subordinate unit.

The documents do not make clear whether the reference to 1st Special Forces Group was a drafting error or an indication of the intended operational customer.

If the reference to 1st Special Forces Group reflects the intended operational customer, the Indo-Pacific context points toward a more specific use case than routine facial recognition.

In that theater, the value of Clearview would likely lie in identity intelligence against gray-zone actors, including China-linked maritime, commercial, militia, intelligence, logistics, or influence networks whose usefulness depends on ambiguity.

Clearview would give analysts and operators a way to take a face from an encounter, video, phone, social post, partner lead, port image, or maritime incident and test whether that person has a wider public footprint.

In an area of operations where the operational problem is often attribution rather than battlefield identification, that capability could help connect identity, affiliation, and pattern-of-life information to people operating below the threshold of open conflict.

Either way, the procurement places Clearview AI inside the intelligence support architecture of Army special operations.

The Clearview justification document says its facial recognition content is “not readily available from any other AI source,” and asserts that the National Institute of Standards and Technology (NIST) had confirmed Clearview’s “superior accuracy and reliability” with a 99.85 percent accuracy rating.

That language should be read carefully. The 99.85 percent figure appears to come from Clearview’s own November 2021 press release, which said NIST testing showed Clearview’s algorithm correctly matched mugshot photos at 99.85 percent in a 12-million-photo sample.

The same Clearview release used the phrase “confirmed the superior accuracy and reliability,” which strongly suggests the Army copied or paraphrased Clearview’s characterization rather than NIST’s own neutral language.

NIST testing should not be read as a government endorsement of Clearview’s deployed platform.

The procurement file also says market research identified only two other sources in use, FaceCheck.ID and PimEyes, but the Army concluded that neither provides the accuracy Clearview does and that only Clearview users can access the system.

The Army specification lists the scope of collection as 50 billion images. Clearview’s current public marketing says its law enforcement database contains more than 70 billion images, which the company describes as sourced from public web sources including news media, mugshot websites, public social media, and other open sources.

The new Army award is notable not only because Special Forces is continuing to use Clearview, but because the sale went through Octaris rather than directly through Clearview.

Octaris is a young company, founded in 2024, but its public profile is built around precisely the type of intelligence work that the Clearview procurement supports.

The company describes itself as a commercial all-source analysis firm that turns complex information into mission-focused intelligence. Its website says it uses proprietary data holdings, publicly available information, commercially available information, and advanced analytical tools to help decision-makers act faster.

Octaris is led by founder Christopher J. Blake, a Navy veteran and former intelligence officer with extensive special operations experience who worked in the private technology and intelligence sector, including with Dataminr’s public sector business.

Octaris’s investor and partner network places the Clearview transaction inside a broader ecosystem of defense technology, intelligence tradecraft, commercial data, and pattern-of-life analysis.

Leonid Capital Partners, Octaris’s first institutional investor, describes Octaris as a veteran-founded defense technology company building SOFOS, an AI-powered system for national security users that unifies fragmented data and automates intelligence workflows. It provides all-source analysis across public, commercial, and proprietary datasets.

In August 2025, Octaris received a roughly $1.2 million Phase II Small Business Innovation Research award for SOFOS, described in public award records as an AI-enabled real-time environment understanding and entity-resolution system for the Department of Defense.

The award runs from August 28, 2025, to February 26, 2027.

The SBIR abstract says Octaris and its development partner, Adelphi Data, are building SOFOS for the Pentagon’s Chief Digital and Artificial Intelligence Office.

Although the contract was routed through an Air Force Life Cycle Management Center contracting office at Wright-Patterson Air Force Base, the stated customer was CDAO, not a local Wright-Patterson base-support function.

CDAO’s mission is departmentwide AI adoption, data fusion, analytic tools, and decision-support capabilities.

The SBIR abstract says SOFOS uses unique identifiers and attributes from publicly available and commercially available data to identify patterns, trends, associations, activities, and indications involving entities of interest to the Department of Defense.

In plain terms, SOFOS appears designed to help analysts connect fragmented data about people, places, organizations, activities, and locations into a more coherent picture of an operating environment.

That is where the Clearview purchase and the SOFOS contract begin to converge conceptually, even though the procurement records do not show that the two systems are formally integrated.

Clearview gives a user the ability to search a face against a massive database of scraped public images. SOFOS, by contrast, is described as a broader entity-resolution and operating environment platform. It is meant to connect people, places, organizations, identifiers, activities, locations, and other attributes across disparate public and commercially available datasets.

For CDAO, the attraction is obvious. CDAO is not merely a contracting office for AI tools. It is the Defense Department office responsible for accelerating AI adoption, data integration, analytic modernization, and decision support capabilities across the department.

A platform like SOFOS would allow CDAO to test how AI can fuse commercial data, open-source intelligence, and other publicly available information into a usable analytic layer for military planners, intelligence analysts, and operational commands.

In practical terms, SOFOS could be used to map a foreign operating environment, identify patterns in adversary activity, connect fragmented identifiers across datasets, surface relationships between people and organizations, analyze movement or behavioral patterns, and help analysts understand how entities relate to a location, function, network, or desired military end state.

That does not mean SOFOS is a targeting system by itself. Nor do the available records show that Clearview’s facial recognition database has been plugged into SOFOS. But the logic of the two procurements is closely aligned.

Clearview helps identify a person from a face. SOFOS is designed to connect identifiers and attributes once a person, place, organization, or activity becomes an entity of interest.

Together, they illustrate the direction of travel inside parts of the defense intelligence ecosystem: biometric search, commercial data, open-source intelligence, and AI-enabled entity resolution are becoming parts of the same operational architecture.

Investor materials say SOFOS is already live in beta with users across multiple Defense Department and Intelligence Community organizations, but no additional public contract awards explicitly for SOFOS were found.

Adelphi, Octaris’s SOFOS partner, adds another layer. The McLean, Virginia-based company publicly describes itself as built by a team with deep experience across national security, intelligence operations, and large-scale data infrastructure.

Its chief executive, Kartik Chopra, has publicly disclosed past work as an intelligence officer in the U.S. Intelligence Community. The company also lists Ralph F. Goff as an advisor. Goff served as CIA Chief of Station in six countries, Chief of Operations for Europe and Eurasia, and Chief of the agency’s National Resources Division.

The National Resources Division is the CIA’s U.S.-based human intelligence component inside the Directorate of Operations. Its mission is foreign intelligence collection conducted from inside the United States.

Adelphi’s ties also extend into the commercial data and advertising technology world. Travis May, a former LiveRamp chief executive and founder of Datavant, is publicly listed as a co-founder and board member of Adelphi.

LiveRamp was previously reported to have sold its cellphone location-tracking business to Cuebiq, another company in the location intelligence ecosystem.

The result is a procurement trail that connects Clearview AI’s facial recognition database, Army Special Forces targeting requirements, a special operations-linked private intelligence firm, a Pentagon AI and entity-resolution platform, former U.S. intelligence personnel, and commercial data infrastructure.

To be clear, there is no indication in the procurement documents that any of the awards involves domestic law enforcement use or bulk surveillance of Americans.

But that absence does not eliminate the broader concern.

Civil liberties advocates have warned that the Pentagon’s expanding use of commercial AI, biometric search, and data-fusion tools raises unresolved questions about how military AI systems could be used or repurposed, especially after controversy over the Pentagon seeking broad access to commercial AI services.

The concern is not theoretical. The Army has its own history of domestic surveillance abuses.

During the 1950s, 60s, and 70s, the U.S. Army Intelligence and Security Command and 902d Military Intelligence Group ran the CONUS Intelligence program, which used roughly 1,500 agents to monitor protest groups and political activity inside the U.S.

Ultimately, the names of nearly 20,000 Americans were entered into computers, dossiers, and files.

The program became part of the broader post-Watergate reckoning over intelligence abuses that led to investigations and modern congressional intelligence oversight by the Senate Select Committee to Study Governmental Operations with Respect to Intelligence Activities.

That history is why civil liberties advocates tend to view any new military access to AI-enabled surveillance and data-fusion tools with legitimate suspicion.

Article Topics

biometric matching  |  biometrics  |  Clearview AI  |  facial recognition  |  identity intelligence  |  U.S. Army  |  USSOCOM