Android Is Finally Reading Your Texts for Scams. For Once, I Can Argue for It.

My verdict: this is one of the few AI features I can name a real use for. Whether you can trust it is a separate question, and the answer is not in yet.

According to Engadget, Google is adding an opt-in setting to Messages that uses Gemini models to judge a sender’s intent and warn you when a message looks like a scam. Gemini Nano also checks links on the device before you tap them. The rollout starts in the coming weeks on the Pixel 10 and later, the Galaxy S26 series, the Galaxy Z Flip and Fold 8, and a few other devices. Two other changes come with it. Later this month, Android 15 and newer will hide passwords, banking details and card security codes by default while you share your screen with Chrome or WebView open. In early December, Android will start blocking apps from switching on call forwarding in the background.

I know several people who were caught by messages that looked like a legal notice or came from what appeared to be their own bank, a copy of the real thing good enough to pass. That is the case where a warning earns its place. A message that looks right to a tired person at the end of a day is exactly what a second pair of eyes is for.

The privacy cost is real, and I will not pretend otherwise. Something has to read the message to judge it. Google says the analysis runs in Private AI Compute, which encrypts the messages and processes them in a space that nobody else, Google included, can access. That is a claim from the company, and I cannot check it. I want to believe it. Until someone outside Google can inspect how it works, I am treating it as a promise and not as a fact.

Note the word opt-in. Nothing here switches on by itself, so someone has to turn it on, and the people most likely to fall for a fake bank message are the least likely to dig through settings. Switch it on for them if you can, and read Google’s own description of how the processing works before you decide.