Google recently announced the deployment of new artificial intelligence (AI) technology to combat online scams on Chrome, Search, and Android platforms.
The company revealed that it will leverage Gemini Nano, an on-device large language model (LLM), to enhance Safe Browsing on Chrome 137 for desktop users.
“The on-device approach offers immediate insights into risky websites, enabling protection against unseen scams. Gemini Nano’s LLM is ideal for this purpose due to its ability to analyze complex websites, facilitating quick adaptation to new scam techniques,” Google stated.
Google mentioned that it is already using AI to combat remote tech support scams, which deceive users into sharing personal or financial information under false pretenses of computer issues.
The AI-driven approach involves using the LLM to assess web pages for potential tech support scam signals, such as the keyboard lock API. Security indicators are then extracted and analyzed by Safe Browsing to identify potential scam pages.
“To ensure the LLM is used sparingly and locally on the device, we manage resource consumption by monitoring token usage, running processes asynchronously, and implementing throttling and quota enforcement mechanisms to control GPU utilization,” explained members of the Google Chrome Security team in a blog post.
 |
| Overview of how on-device LLM assistance in mitigating scams works |
Google plans to extend this feature to detect other scam types, including those related to package tracking and unpaid tolls, with the intention of launching it on Chrome for Android later this year.
In addition, Google has enhanced its AI-powered scam detection systems to identify and block 20 times more deceptive pages from search results, reducing impersonation schemes targeting airlines and official resources like visas and government services by significant percentages in 2024.
Furthermore, Google is introducing a new warnings feature for Chrome on Android that uses on-device machine learning to alert users about suspicious notifications sent by malicious sites, aiming to prevent the download of harmful software or disclosure of sensitive information.
“This new feature utilizes on-device machine learning to detect and warn users about potentially deceptive or spammy notifications, providing an added layer of control over the information displayed on their devices,” explained Chrome Security’s team in a blog post.
Users will see the name of the site sending the notification, a warning message about potentially deceptive content, and options to unsubscribe or view flagged content when a notification is flagged by Chrome.
These developments follow Google’s recent launch of AI-powered scam detection features in the Messages app for Android. The company also introduced measures to identify scam calls last year.
Additionally, Google is reportedly working on an Advanced Protection feature for Android 16, similar to Apple’s security measures, which include disabling JavaScript, 2G connections, and enabling security features like Theft Detection Lock and spam protection by default.
Google is also exploring a feature to detect scams that manipulate users into opening banking apps during phone calls, as reported by Android Authority.
