New information has emerged detailing **Meta's** exploration of advanced face-recognition technology for its smart glasses, sourced from a company deeply embedded in government surveillance. ### The Rank One Computing Connection **Meta** reportedly licensed face-recognition software from **Rank One Computing**, a Denver-based firm deriving approximately 80% of its revenue from government clients. This license was tied to a test version of the **Meta AI app**, which powers **Meta's Ray-Ban** and **Oakley smart glasses**. **Rank One Computing's** technology has a significant footprint in law enforcement and military applications. The **US Marshals Service** utilizes it for prisoner identification, while the **Naval Criminal Investigative Service** (NCIS) purchased its video tool, **ROC Watch**. The company also developed long-range face recognition for **US Special Operations Command**, capable of identifying faces from up to a kilometer away. Police departments nationwide integrate **Rank One's** algorithms into tools from various vendors. ### Unveiling a Covert Collaboration This license is the first known evidence of a direct business relationship between **Meta** and **Rank One Computing**. It provides a rare glimpse into the type of sophisticated technology **Meta** considered for a mass-market consumer device, highlighting the increasingly thin line between surveillance tools for law enforcement and military, and consumer-facing products. The licensed software included **Rank One's** face recognition and liveness detection, designed to differentiate between a live person and a photo or mask. It supports up to 10 million facial templates. Code reviewed by WIRED revealed that remnants of **Rank One's** integration, including routines to load its license and initialize its software, were present in a version of **Meta's** app shipped to millions of consumers this month, albeit dormant, alongside **Meta's** own face-recognition system. ### Dormant Systems and Subsequent Removal Crucially, none of the face-recognition systems linked to **Meta's** smart glasses were ever activated for users. **Meta** subsequently removed all related code from the app on June 5th, a day after WIRED reported on the company's internal, unreleased face-recognition system, codenamed **NameTag**, which was also dormant within the **Meta AI app**. **Meta** has largely remained silent on the specifics of its arrangement with **Rank One Computing**, declining to comment on the reasons for licensing the software, the duration of the relationship, or its current status. **Rank One Computing** also declined to comment. ### Leadership and Operational Reach Founded in 2015 by engineers from the nonprofit research institute **Noblis**, **Rank One Computing** went public on Nasdaq in February. Its leadership boasts deep ties to law enforcement and intelligence agencies. CEO **B. Scott Swann** previously led the **FBI** division overseeing its biometric databases. Its board includes former high-ranking officials from the **CIA**, **FBI**, and the **Pentagon**. **Rank One's** technology is already deployed in various critical applications. The **US Marshals Service** has used a biometric identification kit built on its technology since 2021. In West Virginia, dozens of schools use the software to screen faces against the state's sex-offender registry. Its algorithms are also integrated into products from **DataWorks Plus** and **LexisNexis's Lumen platform**, which assists police in running face searches against state, regional, and **FBI** databases. ### Performance Disparities and Regulatory Gaps Like many face-recognition systems, **Rank One's** technology exhibits demographic performance disparities. Testing by the **National Institute of Standards and Technology** (**NIST**) revealed that a version of the algorithm produced false matches at significantly different rates based on a person's sex and country of birth (used as a proxy for race). Error rates were lowest for individuals born in Eastern Europe and generally higher for women than for men. The U.S. currently lacks comprehensive national regulations governing face recognition. While many states require warrants for police access to such data and are incorporating biometric protections into consumer privacy laws, experts warn of the significant and largely unbounded risks if this powerful technology becomes a common consumer product without proper oversight. Joseph Jerome, a former **Meta Reality Labs** policy official, noted the historical trend of military technologies transitioning into consumer products, describing it as "arguably the story of the internet."