Clearview AI is a US facial recognition vendor founded in 2017 by Hoan Ton-That, headquartered in New York. The company has built a face recognition platform on a database of more than 50 billion images scraped without consent from social media, news sites, and other public web sources. Clearview’s customer base is primarily US federal, state, and local law enforcement, with additional contracts in Ukraine and several other jurisdictions. The company has been the subject of regulatory action and litigation in the US, UK, EU, Australia, and Canada.
What it means in practice
Clearview’s product takes a face image as input and returns matches from its database, each linked to the source URL where the image was originally scraped. Law enforcement use cases include identifying suspects from CCTV footage, identifying victims of human trafficking, and identifying individuals depicted in seized media. The product’s accuracy has been claimed by the company to exceed 99 percent against the NIST FRVT benchmark, though independent verification of operational accuracy in real-world deployment is limited.
Specific things to know
Clearview has accumulated regulatory enforcement actions in multiple jurisdictions: a 22.5 million euro fine in Italy in 2022, GDPR enforcement in France and Greece, the UK Information Commissioner’s Office imposing a 7.5 million pound fine in 2022, and similar enforcement in Australia and Canada. The Illinois Biometric Information Privacy Act litigation produced a settlement in 2022 that prohibits Clearview from selling its database to private US companies. The company’s federal customer base in the US has continued to grow despite international enforcement.
Change today
Anyone who has uploaded a face photo to a public social media account is plausibly in the Clearview database. The operational answer is not to attempt to remove yourself from a database you cannot meaningfully access, but to recognise that face recognition has become an investigative tool with very low marginal cost. For activists, journalists, and others who attend public events, this changes the threat model around routine event photography: a photograph taken at a 2019 protest can be matched against a face captured at a 2025 incident.
Related articles
See our coverage of the Illinois BIPA litigation, the European GDPR enforcement actions against Clearview, and the operational threat model implications of large-scale face recognition for journalists and activists.