Google shared details on how it uses AI-powered tools to fight online scams across platforms like Google Search, Android, and Chrome. Google Chrome's Enhanced Protection mode now uses the Gemini Nano on-device large language model (LLM) on desktop.
Google said in a blog post that the on-device model offers an extra layer of safety against online scams. It provides "instant insight on risky websites and allows us to offer protection, even against scams that haven't been seen before."
Enhanced protection has been available on Chrome since 2020 as part of Safe Browsing. You can enable it by going to Settings > Privacy and Security > Security. The feature offers real-time protection against malicious downloads and extensions, not just websites.
According to the tech giant, tech support scams are among the biggest online threats, and they are being dealt with using the new AI approach. In future updates, Google will expand the security feature to Android devices and more types of scams.
Google will also use on-device machine learning for Chrome on Android to warn unsuspecting users against malicious or misleading website notifications. A warning message will include the option to unsubscribe or view the blocked content.
The tech giant said that investments in AI scam detection systems and improvements in classifiers have enabled it to detect 20 times more scammy pages on Google Search. Over the last three years, it has made several AI-related improvements to its anti-scam systems.
Google uses these capabilities to "analyze vast quantities of text and identify subtle linguistic patterns and thematic connections that might indicate coordinated scam campaigns or emerging fraudulent narratives."
Google noted in a report that it cut down scams by 80% in Search, where scammers posed as airline customer service providers to take advantage of people who needed help. Due to the new protections, misleading pages mimicking official resources like visas or other government services saw a drop of over 70% in 2024.
The report suggests various indicators that can help you catch the scent of bad actors. For instance, they often use lookalike domains to trick people. You should be aware of strange formatting, unusual fonts, or unexpected symbols, and always look for official resources.
0 Comments - Add comment