Google rolls out Gemini scam defenses on Android

Google is adding opt-in Gemini scam detection in Messages, plus call-forward blocks, screen-share masking, and bank-call checks.

Google is rolling out opt-in Enhanced Protection in Google Messages that uses cloud Gemini models to judge scam intent in conversations with unrecognized senders, alongside existing on-device detection. The Verge says that data is encrypted and processed in a dedicated space Google cannot access, while BleepingComputer, Help Net Security, and GBHackers identify the processing as Private AI Compute; GBHackers adds that the assessment covers multi-message social engineering. On-device Gemini Nano inspects unfamiliar links before they open, and BleepingComputer and GBHackers say those checks use web signals. Android will block apps from silently enabling call forwarding and warn before manual or risky MMI and call-forwarding codes take effect, while Chrome and Android WebView will mask login and payment fields during screen sharing—on Android 15 and later, according to Help Net Security. BleepingComputer and GBHackers report that existing detection flags more than 5 million suspected scam calls and texts a month and that verified calling with Banco do Brasil and Itaú Unibanco has blocked more than 3 million spoofed calls since June; Help Net Security instead emphasizes that participating banks, including Banco do Brasil, can automatically end calls that claim to be the bank but fail verification. The Verge also says chat-notification scam detection is expanding from Pixel phones to Honor, Motorola, Oppo, OnePlus, Realme, Tecno, Vivo, Xiaomi, and others, with hundreds of millions more users planned later this year.

• Opt-in Enhanced Protection in Google Messages uses cloud Gemini to assess scam intent in unknown-sender chats; The Verge says data is encrypted in a space Google cannot access, while BleepingComputer, Help Net Security, and GBHackers call…
• On-device Gemini Nano checks unfamiliar links before they open; BleepingComputer and GBHackers say warnings use web or live web signals.
• Android will block apps from silently enabling call forwarding and warn on manual or risky MMI and call-forwarding codes.
• Chrome and Android WebView will hide login and payment fields during screen sharing; Help Net Security specifies Android 15 and later, while The Verge describes concealing sensitive information more generally.
• BleepingComputer and GBHackers say existing detection flags more than 5 million suspected scam calls and texts each month.
• Verified calls with Banco do Brasil and Itaú Unibanco have blocked more than 3 million spoofed calls since June; Help Net Security says participating banks, including Banco do Brasil, can automatically end unverified bank-impersonation…
• The Verge says notification scam detection is expanding from Pixel to Honor, Motorola, Oppo, OnePlus, Realme, Tecno, Vivo, Xiaomi, and others, with hundreds of millions more users planned later this year.