Google deploys AI-powered tools to tackle online scams

Google has unveiled a major expansion of its AI-driven anti-scam initiatives, detailing how advanced tools—powered by its on-device Gemini Nano language model—are helping safeguard users across platforms like Google Search, Chrome, and Android. In a blog post this week, the tech giant emphasized that the fight against increasingly sophisticated online scams is being bolstered by real-time machine learning and language models designed to detect threats proactively—even those not seen before.

A centerpiece of the new strategy is the Enhanced Protection mode in Google Chrome, which now incorporates Gemini Nano, the company’s lightweight, on-device large language model (LLM), for desktop users. This integration allows Chrome to analyze risky websites locally, enabling instant insights and advanced threat detection without relying solely on server-based updates.

Enhanced Protection, originally introduced in 2020 as part of Chrome’s Safe Browsing suite, offers real-time defenses not just against malicious websites, but also against harmful downloads and browser extensions. The new AI layer further increases its ability to detect previously unknown scam tactics, especially those involving tech support fraud—one of the most pervasive threats identified by Google.

According to the company, future updates will bring these AI capabilities to Android devices, expanding their reach and adapting to a wider variety of scams. For example, Google will soon deploy on-device machine learning to help Android users identify and block deceptive website notifications, often used to lure individuals into scams. The notifications will include clear options to unsubscribe or investigate further, giving users more control and clarity.

Beyond Chrome, Google Search is also benefiting from AI advancements. The company revealed that new scam detection classifiers have increased its ability to flag fraudulent websites by 20 times. Over the past three years, enhancements to Google’s anti-scam systems—driven by AI—have dramatically improved detection accuracy and reduced the visibility of coordinated scams.

In one notable success, Google reported an 80% reduction in scam attempts targeting people searching for airline customer service contact details. These scams typically involved fake websites posing as airline support lines. Similarly, misleading pages designed to imitate official government services such as visa applications saw a 70% drop in 2024, thanks to upgraded protections.

Google's AI systems work by analyzing vast quantities of online content to spot subtle linguistic cues and emerging themes that suggest scam operations or misinformation campaigns. The system is also adept at flagging lookalike domains, which mimic legitimate sites using misspellings, odd fonts, or strange characters to trick users.

The company encourages users to stay vigilant by recognizing the signs of fraudulent activity: check for unusual domain names, erratic formatting, and suspicious language. When in doubt, users should seek information directly from official sources.

As scammers continue to evolve their tactics, Google’s AI-driven tools are playing a pivotal role in shaping a more secure online environment. With Gemini Nano now active in Chrome’s Enhanced Protection and more updates on the horizon for Android, Google is setting a new standard in proactive scam prevention.