Google said on its blog that it is rolling out new Android scam defenses. In Google Messages, an opt-in Enhanced protection mode sends chats with unknown senders to Gemini AI models in a private cloud. A smaller on-device model, Gemini Nano, checks unfamiliar links. Android will also blank login and payment fields in Chrome and WebView during screen sharing. It will stop apps from quietly turning on call forwarding. For bank calls, Android asks the bank's app whether it placed the call, and hangs up if not. Since June, Google said, Android has ended more than 3 million fake calls posing as Banco do Brasil and Itaú Unibanco, counting those two banks only.
Those fake calls spoofed bank lines used only for inbound calls, so a call from them is never real. Scammers use call forwarding to reroute a victim's calls and catch one-time passcodes. Android now warns if someone types a forwarding code on the dial pad. Google gave no estimate of how many extra scams the tools will catch. BleepingComputer reported that the Messages mode is off by default. Call-forwarding protection starts in early December.
The strongest feature works only where banks enrol, and the AI layer needs users to switch it on. Bank teams can ask whether their lines are enrolled. Fraud teams can ask how customers will learn the setting exists.