Chrome is getting new scam protections that use Gemini to detect deceptive websites and warn users when they copy suspicious commands from ClickFix attacks.
The changes are part of Enhanced Protection in Safe Browsing, which Google claims makes users ten times safer from phishing, social engineering, notification abuse, and other scams compared with Standard Protection.
Chrome is also adding a new “suspicious site” warning for pages that show deceptive patterns but haven’t reached the confidence level required for its existing full-page red warning.
“Suspicious site warnings notify users when a page exhibits deceptive patterns commonly associated with known malicious sites,” the company explained.
“By tailoring the warning UI in this way, we are able to safely expand our coverage to display a higher volume of protective warnings.”
Gemini is moving from on-device to server-side scam detection
Chrome previously used an on-device Gemini model to detect tech support scams, but it’s now moving to a more powerful server-side model for Enhanced Protection users.
As explained by Google, the model looks at signals such as the brand represented on a page, its intent, and how likely it is to be a scam. Safe Browsing then combines that analysis with domain reputation and other security signals.
“Chrome minimizes the amount of data shared with the server by only analyzing page text when a site exhibits high-confidence scam indicators,” Google noted.
“And whenever page content is sent off the device, it is processed ephemerally – meaning it is never logged, retained or used to train models.”
Chrome can now catch ClickFix commands
Enhanced Protection can also detect when you copy a suspicious command from a ClickFix scam, which typically tricks users into running malicious PowerShell or Command Prompt commands themselves.
If Chrome suspects a ClickFix script, it removes personally identifiable information from the clipboard payload and sends it to Safe Browsing for analysis. If it comes back as dangerous, Chrome displays a warning before you continue.
Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.
The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.
