Google rolls out Gemini scam detection and ClickFix warnings in Chrome

Google said it is rolling out Gemini-powered scam detection to identify deceptive websites in Chrome’s Enhanced Protection Safe Browsing mode, adding warnings for suspicious websites and for commands copied from ClickFix attacks.

All three protections are part of Chrome‘s Safe Browsing Enhanced Protection mode. The new suspicious-site warning tier sits below the existing full-page red dangerous-site warning, which Chrome uses for pages known to be malicious and high risk. Suspicious-site warnings flag pages that show deceptive patterns but fall short of the confidence needed for a red warning, and Google rates a website as suspicious by comparing it with other deceptive or misleading sites. Google says tailoring the warning interface lets it safely expand coverage to show more protective warnings.

Google is moving scam detection to a more powerful server-side Gemini model for Enhanced Protection users, a shift from the on-device Gemini model Chrome used for tech support scams. The server-side model evaluates page content, the brand represented, page intent, and scam likelihood. Safe Browsing combines Gemini output with domain reputation and other security signals for a final verdict, and expanded detection now covers fake antivirus and phony browser update scams in addition to tech support scams. Chrome analyzes page text only when a site exhibits high-confidence scam indicators, which limits the data shared with the server.

Enhanced Protection now warns users when they copy suspicious commands from ClickFix attacks. Safe Browsing processes page content sent off the device ephemerally and never logs, retains, or uses it to train models, Google states. Chrome removes personally identifiable information from the clipboard payload before sending it to Safe Browsing for analysis. If Safe Browsing returns a dangerous verdict for a copied command, Chrome displays a warning before the user pastes the command into a terminal.

ClickFix is a social engineering technique that tricks users into manually copying and running malicious commands to fix a fake issue, an approach that evaded browser defenses because the user performs the action themselves. ClickFix commands often invoke living-off-the-land binaries such as PowerShell and cmd.exe, legitimate preinstalled operating-system tools that attackers repurpose to download malware, hide activity, and abuse the trust that security software places in them. Shifting the download from the browser to the terminal lets attackers bypass Chrome and Safe Browsing download protections, and the copied command can differ from the command shown on the page. Google designed the ClickFix defense to be high precision and to minimize disruption to legitimate developer activity.

Suspicious-site warnings use distinct iconography, colors, and text, and users can still proceed to a flagged site if they are certain it is legitimate. A warning may say the site does not seem right because it resembles sites tied to phishing, malware, and scams. Users can click Learn More, close the warning, click Return to safety, or mark a suspicious site as safe. “Suspicious site warnings notify users when a page exhibits deceptive patterns commonly associated with known malicious sites,” Google said in its blog post.

Users must run Chrome in Enhanced Protection mode to use the new defenses, and Standard Protection does not include them. On desktop, users enable Enhanced Protection through Settings, Privacy and security, Security; on mobile, they use Safe Browsing. Users can also turn on Enhanced Safe Browsing for their Google account through the account page under Security and sign-in, a setting separate from the browser-level option.

Google claims Enhanced Protection makes users ten times safer from phishing, social engineering, notification abuse, and other scams than Standard Protection, a figure based on the average number of safety warnings shown to users of the two modes. Google introduced Enhanced Protection for Chrome Safe Browsing in 2020. Last year Chrome began using a smaller on-device Gemini model for tech support scams, an approach that faced hardware limits, and Chrome requires around 20GB of free space for on-device AI models.

Google confirmed earlier this year that it is using AI to squash over a thousand security bugs in Chrome. Chrome’s new product boss stated the goal is to get people using AI to get things done in Chrome.

ClickFix scams have been widespread lately, and Google describes ClickFix terminal exploits as a rising threat. Google says Chrome and Safe Browsing will keep developing smarter real-time defenses against phishing, scams, social engineering, and malware.