On Thursday, Google announced the launch of new artificial intelligence (AI)-powered countermeasures aimed at combating scams across Chrome, Search, and Android platforms. The company revealed it will start utilizing Gemini Nano, its on-device large language model (LLM), to enhance Safe Browsing in Chrome 137 for desktop users. Google explained, "The on-device approach offers immediate insights into risky websites and enables protection even against previously unseen scams. Gemini Nano's LLM is ideal for this purpose due to its capability to parse the diverse and complex nature of websites, allowing us to respond more rapidly to new scam tactics. " The tech giant also noted that it is currently employing this AI-based method to address remote tech support scams, which frequently attempt to deceive users into divulging personal or financial details by falsely claiming a non-existent computer issue. This system works by analyzing web pages using the LLM to detect signals indicative of tech support scams, such as the use of the keyboard lock API. The identified security signals are then forwarded to Safe Browsing to assess whether the page is likely a scam. Jasika Bawa, Andy Lim, and Xinghui Lu from the Google Chrome Security team commented, "Besides ensuring that the LLM is triggered only sparingly and runs locally on the device, we carefully manage resource usage by monitoring token counts, executing processes asynchronously to prevent browser interruptions, and implementing throttling and quota controls to restrict GPU usage. " Google plans to expand this capability to identify other types of scams, including those involving package tracking and unpaid tolls. This feature is also anticipated to launch on Chrome for Android later this year. Alongside this announcement, Google shared that its AI-powered scam detection systems have been improved to catch 20 times more deceptive pages, blocking such content from search results.

This enhancement has significantly reduced scams impersonating airline customer service by over 80% and those mimicking official resources like visas and government services by more than 70% in 2024. Furthermore, Google is introducing a new warning feature for Chrome on Android that employs an on-device machine learning model to alert users about unwanted notifications sent by malicious websites aimed at tricking them into downloading suspicious software or sharing sensitive information. Chrome Security’s Hannah Buonomo and Sarah Krakowiak Criel explained, "This feature uses on-device machine learning to detect and warn users about potentially deceptive or spammy notifications, giving greater control over the information shown on devices. When Chrome flags a notification, users will see the sender site’s name, a message cautioning that the notification may be deceptive or spammy, and options to unsubscribe from the site or review the flagged content. " These updates arrive just over two months after Google introduced AI-powered scam detection features in the Messages app on Android. Last year, the company also rolled out similar technologies to identify scam calls. The new measures coincide with Google’s preparation of an Advanced Protection feature in Android 16, which in some respects resembles Apple's strategy by disabling JavaScript, turning off 2G connections, and enabling multiple security features by default, including Theft Detection Lock, Offline Device Lock, Android Safe Browsing, and spam protection in Messages.