Scammers stole an estimated $1 trillion worldwide in 2024, according to the Global State of Scams Report. Google says its answer is to catch the fraud earlier, with AI doing the spotting.
By the company’s account, its systems now block 20 times more scam web pages than before. One tactic in particular has shrunk: fake airline customer-service numbers planted in search results, down by more than 80 percent. Protections introduced in 2024 also cut scams that imitate official sites, such as visa portals and government service pages, by more than 70 percent. Google says it keeps hundreds of millions of harmful or scam-related results away from users every day.
Looking for patterns, not single sites
The newer detection tools do not judge pages one by one. AI models read behavioural patterns and linguistic cues to pick out whole campaigns. Google has also trained its systems with large language models to spot a scam in one language and apply the lesson to others.
In Chrome, Enhanced Protection now runs on Gemini Nano, a language model that works on the device and rates the risk of a website as the user loads it. That lets the browser warn about a scam before anyone has reported it or added it to a database. Speed matters here. Phishing sites and fake tech-support pages often appear and vanish within hours.
Phones and messages
Plenty of scams start with a call or text. Google Messages and Phone by Google now run scam detection on the handset, aimed at calls and messages that look harmless at first and turn into fraud later. Chrome on Android also warns users when a site tries to send spammy or misleading notifications, and lets them block or review the alerts.
Google says it has seen scammers work through malvertising, fake travel sites, toll road scams and package-tracking scams.
Sharing the intelligence
Google is a founding partner of the Global Signal Exchange, built with the Global Anti-Scam Alliance (GASA) and the DNS Research Federation, which is meant to speed up the sharing of scam intelligence across borders. In 2024 it also published a report with policy recommendations: better information sharing, incentives for platforms to act early, and more public money for fraud education.
