Google Messages Activates Cloud AI for Deeper Scam Screening
The new Enhanced Protection mode uses Gemini to analyze full conversation context and verify links in real time.
By Lucas Murphy ·

Google has begun rolling out a new security tier for its Android messaging app that relies on cloud-based artificial intelligence to detect fraud. The update introduces Enhanced Protection, which sits above the existing Standard Protection level. This new mode processes data using Gemini models to evaluate the full context of a message thread. The goal is to identify scammer intent more quickly when users communicate with unknown senders.
Google announced the feature earlier this month. It now appears in the latest beta version of the app, identified as version 20261020_00_RC00. The company states that this cloud-based approach handles novel and sophisticated attacks better than the detection tools currently running on devices. The rollout follows a period of testing and development aimed at improving the app's defense against evolving text message scams.
Privacy safeguards are built into the system through Android's Private AI Compute. Messages are encrypted and processed in a secure space. Google describes this space as private to the user, meaning the data is not available to anyone else, including Google. This technical setup allows the system to analyze complex interactions without storing raw chat history in a shared environment.
The update includes enhanced link protection that combines the on-device Gemini Nano model with real-time web signals. This feature checks unfamiliar links before a user taps them. It is available on the Pixel 10 and newer models, the Samsung Galaxy S26 series, the Galaxy Z Flip 8, the Z Fold 8, and other select devices. A third change redesigns warning alerts for links from unsaved numbers, providing more context and time for users to pause.
Users must manually enable the feature. The option is located in the Protection & safety section of the Google Messages settings. The menu now offers three distinct choices: Enhanced protection, Standard protection, and No protection. The app labels the third option as not recommended. The feature is currently available in the latest beta release.
Source: 9to5Google





