Clearview AI's Enterprise Push

What They Said

Clearview AI marketed itself as the most powerful facial recognition tool available — a system that could identify anyone from a single photo by matching it against a database of over 30 billion images scraped from social media, news sites, and public records. The company pitched law enforcement agencies, financial institutions, and retail companies on the technology’s ability to identify suspects, verify customers, and prevent fraud.

What Actually Happened

Clearview’s database was built by scraping billions of photos from Facebook, Instagram, LinkedIn, YouTube, and Venmo — without the knowledge or consent of the people in those photos. When the New York Times exposed the practice in January 2020, the backlash was immediate and global.

Facebook, Google, LinkedIn, Twitter, and YouTube all sent cease-and-desist letters. Australia, the UK, France, Italy, and Greece fined Clearview a combined $70M+ for violating privacy laws. Canada declared the company’s practices illegal. Multiple US states passed or strengthened biometric privacy laws in direct response to Clearview’s practices.

Enterprise clients who had piloted the technology scrambled to distance themselves. Several banks and retailers quietly terminated contracts. The reputational risk of being associated with Clearview’s data practices outweighed any operational benefit the technology provided.

The Root Cause

The data acquisition strategy was the product’s fatal flaw, not a side effect. Clearview’s entire value proposition depended on a database that could only exist through mass privacy violations. There was no consent-based path to building a 30-billion-image facial recognition database. The company’s founders knew this — internal communications revealed they explicitly discussed the legal risks and decided to move forward anyway, betting that the technology’s value to law enforcement would provide political cover.

For the enterprise clients who adopted Clearview, the failure was different: inadequate vendor due diligence. None of the banks or retailers who trialed Clearview asked the fundamental question: “Where did this data come from, and do you have the right to use it?” If they had, the answer would have been disqualifying.

The Pattern to Watch For

Any AI vendor whose competitive advantage depends on data of questionable provenance is a ticking time bomb for enterprise clients. The pattern extends far beyond facial recognition: training data scraped without consent, datasets assembled from sources with ambiguous licensing, models fine-tuned on proprietary data without clear rights chains. If you can’t verify where the training data came from and whether its use is legal in every jurisdiction you operate in, you’re inheriting someone else’s legal liability.

What You Should Steal

Add three questions to your AI vendor evaluation checklist: Where does your training data come from? Do you have documented consent or legal basis for every data source? What happens to my organization’s legal liability if your data sources are challenged in court? If the vendor can’t answer all three with specifics and documentation, the technology isn’t ready for enterprise deployment — regardless of how well it performs.