ai

British Supermarkets Will Test Using Facial Recognition to Verify Alcohol Buyers’ Age, Report Claims

By Melanie Ehrenkranz on at

Select self-service checkouts at British supermarkets will “very soon” be equipped with cameras and facial recognition software, according to the Telegraph. They’ll reportedly use tech from the startup Yoti to determine if a shopper is old enough to buy alcohol and cigarettes without forcing them to “register their identity in advance.”

As it stands, it’s apparently just a pilot scheme, and a mysterious one at that. But the Telegraph reports that self-service checkouts with facial recognition tech could become more widespread next year. Out of the handful of supermarket chains thought to be involved in the pilot, however, so far none have said they are on-board with the reported plan.

According to the report, the goal behind this facial recognition age verification system is to offload the task of checking someone’s ID to the machines. Instead, supermarkets involved will rely on artificial intelligence to estimate the age of a shopper and determine if they can legally purchase certain items. For the pilot Yoti has reportedly partnered with NCR, a US-based company that makes self-service kiosks.

It’s unclear how effective this type of facial recognition software would be. There haven’t been any mainstream success stories, but there have been a number of deeply biased executions of facial recognition technology. There are also some serious privacy concerns, depending on how the companies involved handle this data. But that hasn’t stopped institutions from adopting what appears to be similar technology, so the question is whether supermarkets will buy into the practice, and if the idea will even fly with regulators.

The Telegraph’s report cited three major supermarkets as possible candidates for the tech: Asda, Morrisons, and Tesco. However, ZDNet says Morrisons has “no plans to introduce such technology at this time.” We’ve reached out to Asda, Tesco, and Yoti for more information.

Broadly, it remains to be seen whether this type of software can run smoothly and free from bias. We’ll see. [Telegraph, ZDNet]