Half of Brits struggle to spot AI-driven scams

A National Trading Standards poll of 2,000 UK adults found half of respondents struggle to tell what is genuine online and 25% feel overwhelmed as artificial intelligence makes scams harder to detect.

The watchdog has published an AI Scam Watchlist that describes tools and techniques criminals are using, including cloned voices, deepfake video, AI-generated images and simulated digital humans. The poll found 19% of respondents now avoid some online activity because of scam fears and 31% double-check emails, messages or websites more carefully than before.

The watchlist draws on National Trading Standards intelligence and input from consumer protection partners. It states that AI is not creating new categories of fraud but is amplifying existing scams. Shopping, holiday, investment and romance scams can be personalised, automated and scaled, allowing criminals to reach more people with less technical skill.

National Trading Standards notes that attackers can combine techniques, for example pairing a cloned voice with a convincing website or forged identity, to pressure targets into transferring money or sharing personal data.

The body also flags a likely move toward more autonomous AI agents capable of carrying out complex frauds with minimal human oversight, which could reduce the time and effort needed to run large campaigns and make suspicious activity harder to trace.

Lord Michael Bichard, chair of National Trading Standards, warned that AI is transforming the fraud landscape and enabling organised criminals to produce convincing scams quickly and at scale, and urged government, enforcement agencies and industry to work together to detect and disrupt the networks behind these crimes.

The watchlist includes practical guidance for consumers: verify website addresses, be cautious about unexpected contact that requests money or personal information, confirm offers through official channels before responding and take extra care with messages that push for immediate action.