The Eyes Multiply

Surveillance cameras spread through cities in waves. Each crisis accelerated deployment. London led, driven by IRA bombings in the 1970s through 1990s.

After September 11, 2001, expansion accelerated everywhere. Governments installed cameras at airports, train stations, public squares. Businesses added them for security. Homeowners mounted them on doorsteps.

The cameras multiplied until they became invisible through sheer ubiquity. Part of the urban landscape like streetlights or mailboxes.

For years, these cameras shared a fundamental limitation. Someone had to actually watch the footage. A human needed to sit in front of monitors, scanning for suspicious activity.

Thousands of hours were captured daily. The vast majority never reviewed. The cameras recorded everything but understood nothing. Blind witnesses creating archives that mostly went unseen.

Storage was a challenge. Videotapes filled rooms and warehouses. Digital storage helped but created problems of organization. Finding a specific incident meant knowing when it occurred.

The haystack grew faster than anyone could search for needles. Police complained cameras helped solve crimes after the fact but did little to prevent them.

This limitation began changing with advances in computer vision. Algorithms learned to detect faces. First poorly. Then with increasing accuracy.

By the early 2010s, facial recognition could match photographs against databases of millions. Reliability rates began approaching human performance.

The technology emerged from an unexpected convergence. Facebook trained systems on billions of photos users uploaded. Each tagged friend became training data.

Google developed image recognition through photo services. Apple built face detection into iPhones. The same techniques that tagged your friends could identify strangers on the street.

Commercial development of convenient features created the foundation for comprehensive surveillance.

Governments adopted the technology eagerly. China deployed facial recognition across its surveillance network. The FBI built a database of hundreds of millions of faces.

Police departments acquired systems that could scan crowds in real time. The technology spread from national security agencies to local law enforcement to private security firms.

The Snowden revelations of 2013 exposed digital surveillance at vast scale. Intelligence agencies collected phone records, emails, browsing history from millions.

But this was surveillance of communications. Digital trails left through devices. The new surveillance was different.

> Surveillance of physical presence. Where you go with your body. Who you meet in person. What you do in spaces you thought were anonymous.

Clearview AI demonstrated how far technology could reach. A startup scraped billions of photographs from social media. Built a database larger than any government system.

Law enforcement quietly adopted it. Any photograph could be matched against this database. Not just a name but a complete online history revealed.

> The person next to you on the subway could be identified. Social media accounts. Professional history. Public records. Anonymity in public began to dissolve.

The infrastructure of mass surveillance had been assembled piece by piece. Camera by camera. Algorithm by algorithm.

What remained was to make it truly intelligent. Capable not just of recording and identifying but of analyzing, predicting, alerting in real time. That transformation was already underway.