Google Messages rolls out Enhanced Protection against scams

Google Messages is rolling out an optional Enhanced Protection mode designed to spot more sophisticated scams, using cloud Gemini models through Private AI Compute alongside on-device checks. The update is part of a wider Android security push announced by Google, and it is aimed at conversations and links from people users do not know.¹²

What Enhanced Protection changes

Google says the new mode can use its cloud Gemini models to examine the context of a conversation with an unknown sender. That is meant to catch slow-building scams whose warning signs may be spread across several messages rather than contained in one obvious phishing link. Google Messages already uses on-device detection, so Enhanced Protection adds a cloud-assisted layer for attacks that are harder to classify locally.¹

The cloud processing is routed through Private AI Compute, which Google describes as an encrypted, secure environment that is separate from the company’s ordinary services. The company says messages are processed privately and are not available to Google or anyone else. That is an important part of the pitch: the more capable analysis is supposed to arrive without turning private conversations into a routine data source.¹

Enhanced Protection also upgrades link checking with Gemini Nano. Google says the on-device model can use real-time web signals to judge unfamiliar links before a user opens them, including newly registered scam sites that may not yet appear on conventional blocklists. The mode adds redesigned warnings as well, giving users more context and a longer pause before opening a link sent by a number outside their contacts.¹

A wider Android security rollout

The Google Messages change is only one part of the announcement. Google says Android’s scam-detection tools now flag more than five million suspected scam calls and messages each month, while expanded notification warnings and safeguards against call forwarding and screen sharing are also being introduced.¹ The 9to5Google report says the Messages feature is appearing in the latest beta build and can be enabled under Messages settings, then Protection & safety.²

The setting is opt-in rather than a forced change. Users can choose Enhanced protection, Standard protection or No protection, although Google labels the last option as not recommended. Link-protection support is initially described as covering Pixel 10 and newer devices, the Samsung Galaxy S26 series, the Z Flip 8, the Z Fold 8 and other selected phones.²

Google’s announcement says the rollout will happen over the coming weeks, so availability may depend on the device, software version and beta channel. That also means the feature should not be treated as universally live simply because the announcement has been published. The practical test is whether the Enhanced protection option appears in the user’s own Google Messages settings.¹

Our opinion

This is a sensible direction for scam protection because it targets the part of modern fraud that filters struggle with: context. A suspicious link is useful evidence, but a convincing scam often looks harmless until several messages are considered together. Using a more capable model for that reasoning could improve detection, while Private AI Compute gives Google a clearer privacy argument than simply sending message content to a normal cloud service.

The trade-off is that users still have to trust Google’s explanation of the secure processing environment, and the feature’s usefulness will depend on false positives, battery and data costs, regional availability and how quickly protections react to new scam domains. Enhanced Protection is therefore best understood as another layer rather than a guarantee. It should make suspicious conversations harder to navigate, but the safest response to an unexpected message remains the boring one: pause, verify the sender independently and do not tap first.