Remember trick-or-treating when you were a kid? Going house-to-house in a costume and a mask, with the neighbors trying to guess who you were? We never thought we’d be wearing masks around the neighborhood today (and we have to admit it’s not as fun as we remember). The COVID-19 crisis is raising new questions about identity and privacy. Now that facial recognition technology can pick us out of a crowd—even in a surgical mask—do we really want to be recognized?

Facial recognition (FR) has been around since the 1960s. That early technology was limited to a few features, like the eyes, hairline, mouth and nose. By the late 1990s, cameras and algorithms could identify people with mustaches, beards and glasses. Today’s FR systems are much more sophisticated. Companies like Faception are combining FR with Artificial Intelligence (AI) to match people with various personality traits based on their facial expressions

Of course, the real value of facial recognition lies in its ability to match a face with a name (as well as other personal data, such as an address or customer record). Law enforcement agencies have been doing this for the last 20 years to identify people from surveillance photos and videos, but they have been limited to searching images such as mug shots and driver’s license photos in government databases.

Social media provides a whole new database of face photos to work with. When you upload your image to Facebook, LinkedIn or YouTube, anyone can access it. So, technically, you’re in the public domain. In 2016, a company called Clearview AI figured that might be valuable to law enforcement. So it started scraping photos from every website it could find, then matching them with personally identifiable information. Clearview’s database now contains 3 billion faces (probably including yours). More than 600 law enforcement agencies have licensed Clearview’s software for solving crimes. The company doesn’t make its photo database available to advertisers.

But what if a camera can’t see your face? AI software can also analyze the way we walk. Researchers at the University of Manchester in the UK and the University of Madrid in Spain have developed a system that correctly identifies people by their gait almost 100% of the time. Mastercard^® is working on a payment system for public transport users that’s based on this technology.

For marketers and advertisers, facial recognition and biometric analysis represent a huge opportunity. For the first time, it may be possible to know a person’s identity before they interact with your brand or track a consumer’s interaction with other brands outside of the Internet.

Let’s say you’re in the hotel business. A guest walks up to the check-in desk and their name appears on the receptionist’s computer screen, allowing you to greet them by name even if they’ve never stayed with you before. The same technology can work for self-service kiosks. When a customer approaches, your AI system runs a facial recognition match and pulls the customer’s record from a database, allowing the kiosk to interact with them more like a real person.

Even without access to a facial photo database, smart ads can use facial recognition and AI to promote brands in unique ways. In January 2020, Walgreens tested an interactive system from Cooler Screens for beverage coolers in their Chicago stores. As customers approached the cooler, a camera scanned their faces. Then, a huge LCD screen on the cooler door showed them a customized selection of drinks based on their age, gender and the temperature outside. According to the market test, 80% of shoppers said the system made it easier to find the products they were looking for.

This interactive billboard campaign from General Motors took the idea even further by engaging people in a virtual conversation based on their gender and facial expressions:

How do people feel about facial recognition? It depends. In China, consumers are accustomed to government surveillance. Americans are more wary. In a 2019 Pew Research Center poll, 65% of respondents said they trust U.S. law enforcement agencies to use the technology at least somewhat responsibly. Only 18% trusted advertisers with FR in any situation. In a 2018 Gartner survey conducted in the U.S. and the UK, 52% of respondents weren’t comfortable with AI analyzing their facial expressions in an attempt to understand how they felt, and 63% didn’t want AI to take an “always-on” listening approach to learn more about them.

And, of course, there are privacy issues. In 2019, San Francisco became the first city to ban FR in public places. Four states have passed legislation to restrict the collection of biometric data. Even the Department of Homeland Security has bowed to public pressure and abandoned its plans to use FR on U.S. citizens arriving into and departing the country.

While marketers can still find ways to match faces with names and personal data, there are limits to what certain industries can do with that information. Healthcare organizations, for example, are regulated by the Health Insurance Portability and Accountability Act (HIPAA). They may not use or share personally identifiable information (including biometric data) without consent. Financial organizations are governed by similar regulations. In California, for example, financial information cannot be shared between financial entities, even if they’re affiliated with each other.

Nothing is going to stop the development of more sophisticated technology that can identify people more accurately and match them with their personal information and behaviors. The key to success will be making consumers comfortable with the technology and providing them with real benefits. Remember the results from the Walgreens test? 80% of shoppers said the interactive cooler screen made it easier for them to find the products they were looking for. If marketers can deliver that kind of experience on billboards, websites and interactive TV spots, consumers may feel comfortable showing us who they really are.