Ultra-realistic deepfakes herald the next generation of scams, Which? warns
Deepfake videos created with AI have developed at breakneck speed in recent years, raising concerns about the potential for fraud and the consequences of disinformation. The consumer champion launched an investigation to see how far deepfakes have advanced and whether consumers could tell what is and is not real.
Which? took two AI deepfake videos and two videos with real people in them. A representative sample of the UK adult population (514 people) watched a random selection of three of the four videos and were asked to tell the real footage from the AI-generated fakes. A worrying 70 per cent failed to correctly identify all the real and fake videos shown to them with 8 per cent getting them all wrong and one in five (21 per cent) were only able to correctly identify one.
This is concerning as the research also found that half of those who use social media said they believe they see deepfake videos at least once a day. Which? estimates that this equates to 26 million adults in the UK.
The videos most commonly mentioned in Which?’s survey fell into the entertainment category (60%), while 41 per cent reported the fake videos they saw were political, 29 per cent saw healthcare videos and 25 per cent saw financial content. Disinformation in areas that affect vital life decisions such as these can be dangerous.
The consumer champion also asked specialist researchers from Kingston University’s deepfake detection team to assist in sorting fact from fiction.
The Kingston University team used a variety of deepfake detection tools to analyse a set of fake videos sent by Which?. Surprisingly, only three out of eight videos came back as conclusively fake.
A separate expert, who only had time to analyse three out of the eight videos, used their own detection algorithms and concluded that all three were fake. However, those three videos had been identified as real by the team at Kingston University - further underlining how difficult it is even for experts with cutting-edge tools at their disposal to spot the fakes.
Most of the videos Which? found were adverts placed on social media sites, meaning that the platforms they were posted on profited from them. This is problematic because Which? investigations have found time and again that major social media platforms are teeming with scam ads.
Juniper Research estimates that social media companies made nearly £3.8 billion in 2025 from scam ads in Europe alone. Meta is also being sued in the US by the Consumer Federation of America for failing to protect users from scam ads.
Earlier this year, the UK Government published its long-awaited fraud strategy, outlining its plans to tackle the country’s biggest crime. However, Which? argues that the proposals will fail to stop scams reaching people in the first place and falls short of what’s needed to solve the UK’s fraud epidemic.
Which? believes that the most effective way to disrupt fraudsters would be to force tech giants to take responsibility for the flood of scams on their platforms. The Government needs to make Ofcom take action against tech firms that fail to block scams on their platforms by urgently implementing robust codes of practice for the Online Safety Act’s measures on fraudulent advertising.
Rocio Concha, Which? Head of Policy and Advocacy, said:
“Our investigation into deepfakes on social media shows the public - and even experts - are increasingly at the mercy of scammers exploiting inadequate security on social media platforms.
“This is troubling as we believe the tech giants who profit from scams on their platforms will not take enough action against fraudsters unless they are legally compelled to do so.
“Ofcom must stop kicking the can down the road and implement the Online Safety Act’s measures on fraudulent advertising as soon as possible and the Government must ensure that it does so urgently and robustly.”
-ENDS-
Notes to Editors
Methodology:
Which? surveyed 514 adults in the UK between 6 and 9 March 2026. Fieldwork was carried out online by Deltapoll and the final data has been weighted to be representative of the UK population aged 18+. Each participant saw three videos out of a poll of four (two real, two fake).
“Juniper Research estimates that social media companies made nearly £3.8 billion in revenue from scam ads in 2025 alone…” Source: Juniper Research whitepaper commissioned by Revolut (published February 2026).
Advice for consumers:
Spotting deepfake videos with the naked eye is increasingly difficult, but there are certain ‘tells’ which indicate that AI was involved:
Distortion: In one of the videos that was tested there was visible distortion around the speaker’s hands which also had physically unrealistic proportions.
No eye contact: AI-generated humans often struggle to look directly at the camera.
Too perfect: Faces can be overly smooth and lacking in normal skin features.
Changing details: AI struggles to keep small details perfectly consistent - close examination of physical features or clothing can reveal impossible changes happening moment-to-moment.
Pictured: A deepfake ‘health influencer’ who struggles to look directly into the camera when speaking.
Pictured: A deepfake featuring a woman with noticeably distorted and unrealistic hands. The Boots logo on her uniform undergoes subtle changes over the course of the video.
Right of Replies:
Meta removed the videos shared by Which?.
A spokesperson for Boots said: “We are aware of the circulation of advertisements for an unlicensed weight loss product on Facebook that appear to be from Boots. These adverts are fake and we urge customers to be vigilant. We proactively monitor for this type of content on social media and report any we identify to the platform owners and request that they are urgently removed.
About Which?
Which? is the UK’s consumer champion, empowering people to make confident choices and demand better. Through our research, investigations and product testing, we provide trusted insight and expert recommendations on the issues that matter most to consumers.
Fiercely independent, we put people over profit - shining a light on unfair practices, influencing policy and holding businesses to account to make life simpler, fairer and safer for everyone.
The information in this press release is for editorial use by journalists and media outlets only. Any business seeking to reproduce information in this release should contact the Which? Endorsement Scheme team at endorsementscheme@which.co.uk.