Digital crooks change tactics, Google changes strategy. With Gemini Nano, AI arrives directly on your devices to track the scams even before they act.
Google has just taken a new step in The fight against online scams. The web giant deploys AI protections directly integrated into the devices, on Chrome and Android. The model Gemini Nanolight version of its large language model, analyzes the suspicious pages on the fly. “It helps us adapt to new scam tactics”specifies Google. AI detects subtle signals, even on sites never seen before, and reacts in real time.
In Chrome 137 for computerssecure navigation now benefits from this on -board AI. The analysis is done locally for more responsiveness, without using the cloud. The system assesses the pages using signals such as the abusive use of the keyboard locking API. The results are then transmitted to the secure navigation service of Google, which blocks the suspect sites. The execution is done in the background, with fine management of system resources to avoid any slowdown.
A first success against scams with technical support
Google already uses this technology to counter scams related to remote technical assistance. These frauds push the victims to disclose their personal or banking information pretending to be a breakdown. Thanks to the Gemini Nano model, these pages are analyzed and blocked before even a click is made.
Jasika Bawa, Andy Lim And Xinghui Lumembers of the Chrome Safety Team, specify: “We limit the consumption of GPU and execute everything in a asynchronous way. »»
As of this year, Google will extend the functionality to other current scams. In particular are targeted the false follow -ups of packages and the scams with unpaid toll fines. These practices are multiplying, often by SMS or via intrusive notifications. In response, an AA Alert Integrated with Chrome on Android will be able to block suspicious requests. A local automatic learning model will judge notifications and warn the user in real time.
Results already very telling in research
In parallel, Google has updated its IA filters on Search to detect twenty times more misleading pages. The search engine thus blocks more than 80 % of sites imitating customer service of airlines. It also neutralizes any more 70 % of scams pretending to be official services like government visas or portals. A dynamic that confirms the effectiveness of constant monitoring controlled by artificial intelligence.
Reinforced safety vision on Android 16
Google also prepares Android 16 with measures inspired by Applesuch as the deactivation of JavaScript or 2G connections. The next version will include default protections: locking after flight, protected navigation and spam control in messages. The company even tests to detect calls encouraging to open bank applications. The objective is clear: protect the user when attack occurswithout external intervention.
.