A recent study by tech retailer Currys looked at Google searches in the UK around scam emails, and it seems Evri is the favourite brand scammers are using to target their victims, with 39,978 searches over the past year for “evri scam email”. These searches have also increased by 823% over the past 12 months, when comparing search volume in March 2023 to February 2024.
The next brand scammers hide behind most is PayPal with 28,900 searches over 12 months for “paypal scam email”, however, the monthly search volume for this has decreased by 19% over the past 12 months. And Amazon was the third brand most people were searching in relation to scam emails, with 14,000 annual searches for “amazon scam email”.
AI can’t detect 10% of scam emails
The study also used two forms of AI to analyse a mixture of 40 genuine and scam emails, asking it to identify whether the email was a phishing attempt or a genuine communication from a brand.
The study discovered that both AI assistants incorrectly identified 10% of the mixture of scam and legitimate emails fed to them. Perplexity was better at discovering the spam emails, correct in 28 out of 29 cases, compared to Chat GPT 4 which only identified 27 out of 29 correctly.
And when it came to the legitimate emails, both recognised even less. Perplexity incorrectly categorised three genuine emails as spam and Chat GPT 4 did the same for two.
This highlights the increased sophistication of phishing emails, suggesting more of the public will fall victim to them, particularly as the National Cyber Security Centre (NCSC) said that AI would “almost certainly” increase the volume of cyber-attacks and heighten their impact over the next two years.
This is due to cyber criminals using the technology to approach potential victims in a more convincing way by creating fake “lure documents” (such as scam emails) that don’t contain the translation, spelling or grammatical errors that tended to give away phishing attacks.