Google Rolls Out AI Scam Defenses for Android Messages and Calls

Google Expands AI-Powered Scam Protections Across Android

Google announced a new round of AI-powered defenses against messaging scams and fraudulent phone activity on Android, rolling out to eligible devices starting October 15, 2026. The update layers Gemini-backed analysis onto Google Messages and the Phone app, adds protections against silent call-forwarding abuse, expands verified bank calls, and introduces sensitive-data masking during screen sharing. Google said mobile scams have "evolved far beyond basic robocalls and phishing texts, turning into sophisticated operations that use spoofed numbers, fake bank alerts, and AI voice cloning," adding that "with scams growing harder to spot, it's nearly impossible for users to catch every red flag themselves." The company says its existing scam-detection tools in Phone and Messages already flag more than 5 million suspected scam calls and texts every month.

Details

How the New Protections Work

Enhanced Protection in Google Messages

The headline feature is an opt-in Enhanced Protection mode in Google Messages that scrutinizes conversations with unknown senders for fraud signals. For complex or ambiguous conversations, Messages sends context to a cloud-based Gemini model running inside Private AI Compute, Google's encrypted processing environment designed to analyze sensitive content without exposing it to human review or storing it outside the secure enclave. For simpler checks — such as evaluating an unfamiliar link — the on-device Gemini Nano model cross-references real-time web signals locally, without sending the message content to the cloud at all. Google has also redesigned the warning shown before a user taps a link from an unsaved contact, making the risk clearer before the page opens. Because the cloud-based analysis is disabled by default, users must manually enable Enhanced Protection to benefit from it.

Call-Forwarding Abuse Blocking

A common scam tactic involves tricking a victim into enabling call forwarding — either through a malicious app running silently in the background or by having the victim dial a Man-Machine Interface (MMI) code on the keypad — so that incoming calls (including bank verification calls) are secretly redirected to an attacker. Android will now block apps from silently activating call forwarding without explicit user interaction, and will surface a warning before a manually dialed MMI forwarding code takes effect. Google says this protection begins rolling out to Pixel and "other select devices" in early December 2026.

Expanded Verified Bank Calls

Google's verified bank calls system, which checks whether an incoming call genuinely originates from a bank's call center and automatically terminates calls that fail verification, is expanding to more financial institutions globally, including Revolut. Google credits the system with blocking more than 3 million fraudulent calls impersonating Banco do Brasil and Itaú Unibanco since its June 2026 launch in Brazil — a market where bank-impersonation scams reportedly surged 195.7% over the prior twelve months.

Screen-Sharing Data Masking

To counter "remote assistance" scams in which a caller convinces a victim to share their screen and then reads sensitive information off it, Chrome and Android WebView will automatically blank login fields and payment screens during live screen-share sessions. This protection is rolling out this month to devices running Android 15 and later.

Wider Rollout of Existing Scam Alerts

Separately, Google is extending its existing notification-based scam alerts in Messages to "hundreds of millions" more users through device-maker partnerships with HONOR, Infinix, iQOO, Motorola, OPPO, OnePlus, and Xiaomi, expected later in 2026.

Impact Assessment

Recommendations

For Android Users

• Update Google Messages and the Phone app to the latest version and manually enable Enhanced Protection in Messages settings, since the cloud-based scam analysis is off by default.
• Treat unsolicited links from unsaved contacts with caution even with these protections enabled; no automated filter catches every scam.
• Avoid sharing your screen with unknown callers or "support" contacts, and verify any request to enable call forwarding before confirming it.
• Check with your bank directly if you receive an unexpected call about account issues, rather than acting on instructions given during the call itself.

For IT and Security Teams

• Communicate the Enhanced Protection opt-in step to end users and family members on managed or BYOD Android fleets — the feature provides no benefit unless activated.
• Track device eligibility across your fleet; flagship-only features (Gemini Nano checks) mean a meaningful share of deployed Android hardware won't see the strongest protections at launch.
• Incorporate call-forwarding abuse and remote-screen-share scams into user security-awareness training, since these remain viable attack paths on devices that haven't yet received the update.

For Admins Managing Shared or Loaner Devices

• Confirm the Android version on managed devices; screen-sharing data masking requires Android 15 or later.
• Where possible, standardize on devices expected to receive call-forwarding protections in the early rollout window (Pixel and select partners) for staff handling sensitive communications.

Key Takeaways

1. Google is layering Gemini-based AI analysis — both on-device (Gemini Nano) and cloud-based (via Private AI Compute) — onto Android Messages to catch sophisticated, conversational scams that static filters miss.
2. The strongest cloud-based protection, Enhanced Protection in Messages, is opt-in and disabled by default, so users must manually turn it on.
3. New call-forwarding abuse blocking targets a known technique for hijacking bank verification calls and 2FA, rolling out to Pixel and select devices starting early December 2026.
4. Verified bank calls have already blocked over 3 million fraudulent calls impersonating two Brazilian banks since June 2026, and the system is expanding to more institutions, including Revolut.
5. Screen-sharing data masking in Chrome and WebView, rolling out this month to Android 15+ devices, blanks sensitive fields during remote-assistance scams.
6. The most advanced on-device AI checks are currently limited to flagship hardware (Pixel 10, Galaxy S26, Z Flip7/Fold7), while broader notification-based scam alerts reach other OEM devices later in 2026 — meaning protection levels will vary significantly across the Android ecosystem for some time.

Sources

• Android gets AI defenses against messaging scams, fake calls — BleepingComputer
• New Android protections to help keep you safe from mobile scams — Google