Mind Matters Natural and Artificial Intelligence News and Analysis
Security camera on modern building. Professional surveillance cameras. CCTV on the wall in the city. Security system, technology. Video equipment for safety system area control outdoor. AI generated.
Image licensed via Adobe Stock

When Video Surveillance Gets It Wrong

An incident in Argentina calls the country's facial recognition tech into serious question

A new in-depth report from Wired recounts the story of an Argentinian man, Guillermo Ibarrola, who was falsely accused of a crime and pinpointed via video surveillance. According to Wired, Ibarrola spent nearly a week in custody before the “mixup” was spotted and authorities realized they had nabbed the wrong Guillermo Ibarrola. It was a data entry mistake.

The case is only one of many others in Argentina’s mishaps with video surveillance, and how data mistakes can end up impacting innocent lives significantly. Karen Naundorf writes,

When speaking of South America, mass surveillance technology is likely not the first thing that comes to mind. But a study by the data protection organization Access Now shows Argentina is one of the most surveilled countries in the region, along with Brazil and Ecuador. There are more than 15,000 surveillance cameras in Buenos Aires alone. Facial recognition systems also are in use in the cities of Mendoza, Córdoba, Salta, San Juan, Tigre, and San Salvador de Jujuy. While US cities like San Francisco and Boston have banned real-time facial recognition in public spaces, South America is investing. Critics see this as a worst-case scenario: The technology is being used without an adequate regulatory framework and sufficient controls.

-Karen Naundorf, The Twisted Eye in the Sky Over Buenos Aires | WIRED

While a typical citizen in the city of Buenos Aires will list inflation and crime as higher concerns than privacy violations imposed by the thousands of surveillance cameras in the city, it remains a “scandal” that the technology has so little oversight.

AI facial recognition technology has had its fair share of problems too in misidentifying people. While theoretically, we could claim that an algorithm might be more objective and successful than humans at such tasks, more needs to be discussed. For instance, AI can’t understand context, nuance, or other details that might lend insight to a legal case. This is something lawyer Richard W. Stevens has written about in the past. There are limits to what AI can do. Making one data mistake, these days, could mean an innocent man in jail.

Of course, the issue of privacy writ large is also highlighted in the Wired article. While we might feel safer with security cams everywhere, such tech does make countries more vulnerable to government control and overreach. When mistakes like this are made, and privacy is more or less an illusion of the past, is “safety” worth it?

For further reading: Face recognition: Is the U.S. Copying China’s Surveillance State? | Mind Matters

Mind Matters News

Breaking and noteworthy news from the exciting world of natural and artificial intelligence at MindMatters.ai.

When Video Surveillance Gets It Wrong