Google has announced that it is deploying powerful AI tools across Search, Chrome and Android to block fraudulent content and protect users from evolving scam tactics.
Why?
Online scams are nothing new. However, their scale, sophistication and impact are growing fast. From fake airline customer service numbers to dodgy tech support pop-ups, cybercriminals are increasingly exploiting trust, urgency, and confusion. Now, Google says it’s fighting back with a suite of AI-powered tools aimed at spotting scammy content before users even see it.
“We’ve observed a significant increase in bad actors impersonating legitimate services,” the company stated in its latest blog update. “These threats are more coordinated and convincing than ever.” According to Google, its upgraded detection systems are now blocking hundreds of millions of scam-related search results every day, 20 times more than before, thanks to recent AI upgrades.
What Google Is Actually Doing
At the centre of Google’s push is its latest generation of AI models, including Gemini Nano, a lightweight version of its flagship large language model (LLM), designed to run locally on users’ devices.
Google says it’s deploying the AI toolkit in the following ways:
– In Search. AI-enhanced classifiers can now detect and block scammy pages with significantly higher accuracy, particularly those tied to impersonation scams. A key focus is identifying coordinated scam campaigns, such as fake airline or bank helplines, which Google says it’s reduced by more than 80 per cent in search results.
– In Chrome (desktop). Gemini Nano is being used in Enhanced Protection mode, offering a more intelligent layer of scam detection by analysing page content in real time, even if the threat hasn’t been encountered before.
– In Chrome (Android). A new machine learning model flags scammy push notifications, giving users the option to unsubscribe or override the warning. This is a direct response to the trend of malicious websites bombarding mobile users with misleading messages.
– In Messages and phone apps. On-device AI is scanning incoming texts and calls for signs of scam activity, aiming to intercept deceptive social engineering attempts before users fall victim.
Shift To On-Device AI
The shift towards on-device AI is a critical part of Google’s strategy because, rather than relying solely on cloud processing, running models like Gemini Nano locally means decisions are faster, more private, and can spot never-before-seen scam tactics in the moment.
Why This Matters for Users
For everyday users, the benefits are likely to be fewer scammy links in search results, smarter filters on your phone, and more proactive browser protections.
For businesses (especially SMEs) the impact could be even more significant. For example, according to UK Finance, authorised push payment (APP) scams targeting consumers and businesses cost victims over £485 million in 2022 alone. Many of these start with a search result, a fake email, or a deceptive phone call. However, having Google’s AI defences in place could mean:
– Staff are less likely to stumble on phishing sites during routine searches.
– Malicious browser notifications can be flagged before they cause confusion.
– Company phones and SMS channels are better protected from social engineering attempts.
These AI tools essentially reduce the attack surface for fraudsters, which is clearly an especially valuable outcome for over-stretched IT teams trying to keep up with threats.
What’s in It for Google?
Although Google’s broader rollout of scam-fighting AI may be good for PR, it’s now really a business necessity. This is because public trust in online services, especially search engines and browsers, largely depends on keeping scam content out.
Google is also keen to differentiate itself from rivals like Microsoft and Apple. For example, Microsoft Edge and Bing also use AI to detect malware, phishing and fake websites. Apple’s latest iOS versions include some machine learning-driven protections for spam and scams in Messages and Mail.
Google appears to be going further by embedding AI defences across all major entry points, i.e. Search, Chrome, Android, and communication tools. That integration could give it an edge, especially in markets like the UK where Android holds a dominant share of mobile devices.
However, there’s a potential catch. As AI becomes central to scam detection, the bar will rise for other tech companies too. Users may start to expect this level of protection as standard, which means any platform not keeping up could find itself falling behind in both security and credibility.
Real-World Examples
Google’s own data shows the power of its AI-driven changes. A sharp rise in airline impersonation scams was swiftly countered by enhanced detection models, reducing exposure by over 80 per cent. These scams typically lure users searching for flight changes or refunds into calling fraudulent hotlines, where they’re pressured into handing over personal or financial information.
Another major focus is remote tech support scams, where a pop-up warns users of “critical issues” and urges them to call a fake number. Google says that Gemini Nano can now analyse these deceptive pages in real time, warning Chrome users before they take the bait.
The on-device models also mean that even zero-day scam campaigns (those not yet logged in Google’s vast threat database) can still be intercepted by identifying linguistic and structural red flags.
Room for Improvement?
While the rollout of AI-based protections has been welcomed by many, it’s not without its challenges.
One concern is transparency. AI models can be difficult to audit, and users may not always understand why a particular site or message was flagged. Google says it allows users to override warnings and give feedback, but questions remain about how this data is used and whether false positives could impact legitimate content.
There’s also the issue of resource disparity. Large tech firms like Google and Microsoft can afford to train massive language models and deploy them globally. However, smaller competitors, privacy-focused browsers, or regional search tools may struggle to match these protections, thereby potentially creating a security gap.
Finally, there’s a sustainability angle to consider. Running large AI models, even ones optimised for on-device use, carries an environmental footprint. Google has committed to net-zero emissions by 2030, and claims its Gemini models are designed for efficiency. But watchdogs may still press the company to show how its AI-driven safety tools align with its green ambitions.
What Does This Mean For Your Business?
From a user perspective, integrating scam detection into the everyday tools people rely on may help close the gap with the scammers who often seem to be one step ahead. The use of LLMs like Gemini Nano should mean that Google can now respond faster, spot patterns earlier, and intervene more precisely, whether it’s a fake support call, a misleading notification, or a deceptive search result.
For UK businesses, particularly SMEs without dedicated cyber teams, this could offer much-needed support. With employees less likely to fall foul of phishing links, fake helpdesk numbers, or scammy browser alerts, the business case for Google’s AI defences is strong. It could also lessen the reputational and financial risks posed by impersonation scams, which is a problem that has hit sectors from travel to retail and beyond. That said, relying on a single tech platform for frontline defence carries its own risks, making it all the more important for firms to combine these tools with their own cyber awareness and training efforts.
At the same time, Google’s move is likely to put pressure on its competitors to keep pace. AI-driven scam detection is rapidly becoming a baseline expectation, not a luxury feature. While Apple and Microsoft are investing in their own protections, they may need to match Google’s scale and cross-platform integration to stay competitive, especially as consumer and regulatory expectations around online safety continue to rise. Whether others will follow suit with the same breadth and transparency remains to be seen.
That said, despite the progress, it’s clear that AI alone won’t fix everything. Transparency, accountability, and environmental responsibility all remain live concerns, especially as these systems scale.
