National Trading Standards launches AI scam watch list as deepfake fraud fears grow

National Trading Standards has launched an AI scam watch list as concerns grow about deepfake fraud. New figures published today by the body suggest nearly half of UK adults say it is getting harder to tell what is genuine online. It also found that 79% of people are concerned about a friend or relative becoming a victim of an AI-enabled scam, while 27% are worried about becoming a victim themselves.

Trading Standards said the watch list is intended to protect people from digital humans and AI images online. The move comes as scammers use increasingly sophisticated tools to create fake adverts, fake voices and fake images that can look and sound real.

A consumer expert from Which? said the watch list is a good idea and shows some action is being taken. But Rob Lilly-Jones said the main concern is the volume of scams appearing every day. Thousands pop up and are hard to track, he said. He called on big tech firms to take the problem more seriously and remove scams as soon as they appear, noting that some remain live for days, weeks or even months.

Which? showed examples of the kind of fraud it is tracking. One appeared to use AI to repurpose footage to look like a supermarket advert, in a style the expert described as a classic deepfake scam used to sell fake products. Another used an influencer's likeness to sell a product without their involvement. The expert said deepfakes are one of the most popular AI scams, alongside voice cloning. Fake days out are another emerging trick, with scammers selling cheap tickets to events that do not exist and sending victims to fake websites.

The figures suggest public concern is uneven. Only 27% of UK adults are worried about becoming a scam victim themselves, even though a much larger share fear for friends or relatives. The expert said the technology is getting harder to spot and filters on social media can make checks more difficult.

For people trying to spot an AI fake, the advice includes looking at the eyes, whether they make direct eye contact, the side of the face and the hands. AI images can still show too many fingers, such as six or seven. A glossy or shiny look can also be a clue, though the expert warned that filters make this an imperfect test. Staying offline is probably the safest option, he said, but it is not realistic for most people.