Google Messages is now more secure

Advanced protection added to Google Messages scans all correspondence you have with someone you do not know with Gemini. The setting is optional and does not turn on automatically. Chats are analyzed on the cloud side; The place where your data is processed also changes.

According to a Google survey conducted with YouGov in the US, Canada and Australia, 40% of smartphone users are very or extremely concerned about mobile fraud. Conversations that last for days, starting with cargo information, a job offer or a simple hello, are the main cause of this anxiety.

What does advanced protection do?

Scam Detection in Messages is already working. According to Google, the feature flags more than 5 million suspicious calls and messages in the Messages and Phone app every month. Standard control is on by default, but only covers correspondence with those not in your contact list. On the phone, it works with smaller models.

Advanced protection adds two things on top of this. First, when you're corresponding with someone who isn't in your contacts, Google's cloud-based Gemini models can assess whether the sender is trying to scam you.

Second, a smarter connection control. Running on the phone, Gemini Nano scores unfamiliar connections using live web signals. This helps catch new scam sites that have not yet made it onto the blocklists.

A renewed warning screen also appears before opening a connection from an unregistered number. This screen gives clearer information about the connection. The warning was previously only available in some countries or on phones with Android's Advanced Protection mode turned on. Google announced the feature along with other Android fraud measures on October 14, 2026; It started coming to the phones later in the same month.

Privacy: Private AI Compute

The analysis is done on Private AI Compute, the secure cloud environment that Google uses for its large Gemini models. Google says that the data is encrypted here and no one, including Google, can access it.

Scope is limited: only correspondence with senders you do not know is reviewed. Conversations with family and friends are left out.

If you report a message as spam, the process works differently. Google sends the reported message and its content to itself for analysis outside of Private AI Compute.

How to open?

Open Google Messages and tap your profile picture in the top right.

Enter the Protection & safety section.

Under Spam and scam protection, select Advanced protection.

There are Advanced protection, Standard protection and No protection options on the screen; Google does not recommend the last one. The standard option preserves the on-device spam filter and known malicious link checks. If you change your mind, you can go back with the same steps.

Cloud fraud analysis supports Google Pixel 6 and later, Samsung Galaxy S24 series and later. Models from Galaxy Z Flip 6 and Z Fold 6 onwards and other Android AICore devices are also on the list.