Google expands Android AI defences against fake calls, phishing and call diversion

Google is widening Android’s fraud protections across messages, calls and screen sharing, using AI to detect suspicious links, impersonation attempts and attempts to redirect users’ phone calls.

Google is expanding Android’s defences against scams and impersonation attacks as phone users face increasingly sophisticated fraud attempts that combine fake caller IDs, AI-generated voices and long-running social engineering campaigns. In a statement highlighted by t3n, Google said its AI-driven fraud detection already identifies more than five million suspected scam calls and messages each month, and the company is now widening protections across messaging, calling and screen-sharing features. The push reflects a broader effort to make Android less vulnerable to tactics aimed at stealing login details, banking credentials and other sensitive information.

One of the main changes is an upgraded protective mode in Google Messages. When it is switched on, the system can use larger Gemini models in the cloud to analyse more complex conversations from unknown senders, while Gemini Nano can inspect new links on the device in real time, including suspicious web addresses that have not yet appeared in conventional phishing lists. Google says the cloud processing is encrypted through Private AI Compute so that the content cannot be accessed. The company is also revising warning prompts when users open unfamiliar links from senders not saved in their contacts, and it plans to extend notification-based scam detection, first introduced on Pixel phones, to devices from other manufacturers later this year.

Google is also tightening protections around screen sharing and call forwarding, two techniques scammers often exploit to reach private data. Android already hides confidential information inside native apps during screen sharing, and Google is extending that safeguard to Chrome and Android WebView, the component many apps use to display web content. The system is also being updated to block apps from enabling call forwarding in the background. If a user is prompted to enter a manual forwarding code on the dial pad, Android will now warn them before the code takes effect, a move aimed at stopping attackers who try to divert calls to a number they control.

The company is further developing anti-spoofing checks for calls that appear to come from banks. According to Google, Android is working with official banking apps to examine incoming suspicious calls in real time and verify whether the bank actually initiated the contact. If the caller fails that test, the call can be ended automatically and the user notified. Google said Android has already blocked more than three million fake calls since June from Banco do Brasil and Itaú Unibanco alone, with Revolut and several large banks also among its partners. Android Central has separately reported that Google is rolling out additional theft-protection tools, including motion-based detection, an offline lock and a remote lock feature, while other recent security additions include live threat detection in Google Play Protect and new safeguards for shared screens.