Google deploys AI to fight scams on Android with real-time detection

Google is deploying AI to fight AI-powered scams on Android devices, introducing sophisticated real-time detection systems for both text messages and phone calls. These new features represent a significant advancement in mobile security, using on-device artificial intelligence to identify conversations that start innocuously but later develop suspicious patterns—a common tactic used by modern scammers. By implementing contextual analysis rather than simple filtering, Google aims to counter increasingly sophisticated social engineering attacks that traditional security measures often miss.

The big picture: Google has announced two AI-powered scam detection features for Android that analyze communications in real-time to protect users from sophisticated fraud attempts.

• The company is specifically targeting scams that begin with seemingly harmless interactions but gradually escalate into fraudulent schemes.
• Both features use on-device AI processing to maintain privacy while offering protection against evolving threats.

Key details about text message protection: The scam detection for Messages feature enhances existing spam protection by analyzing entire conversations for suspicious patterns.

• When the AI detects potential scam activity in a text thread, it warns the user and offers options to dismiss the alert or report and block the sender.
• The system works with SMS, MMS, and RCS messages, and is enabled by default for conversations with people not in the user’s contact list.
• Google promises privacy protection, stating that only sender details and recent messages are shared if a user chooses to report suspicious content.

Current availability: The text message scam detection is now rolling out in English across select regions with plans for expansion.

• The feature is currently available in the US, UK, and Canada.
• Google intends to bring the protection to additional countries in the future.

Phone call protection details: The phone call scam detection system, initially launched as a beta for Pixel 9 users, identifies potential scams during live calls.

• When the system detects suspicious patterns, such as requests for gift card payments, it alerts users through audio, haptic, and on-screen notifications.
• Unlike the text message feature, call scam detection is disabled by default and requires manual activation.
• Testing revealed that devices with the more advanced Gemini Nano AI (Pixel 9) outperformed phones using smaller machine learning models (Pixel 6).

Implementation differences: The two scam detection features differ in their default settings and availability.

• Text message protection is automatically enabled, while call protection requires manual activation.
• Call protection remains limited to Pixel 9 users in the US, though Google is expanding the beta to all Pixel 9 owners.
• Both features prioritize privacy by exclusively processing data on-device and not storing or transmitting conversation content.