- Google Chrome is testing a new AI-powered scam detection feature
- It seemingly uses an on-device Large Language Model (LLM) to maintain user privacy
- AI-driven safety tools, including scam detection, help to fight the rise of AI-powered threats online
The world’s most popular browser, Google Chrome, is experimenting with a new AI-powered tool designed to help you avoid online scams.
The feature is currently being tested and apparently uses AI tech, specifically a Large Language Model (LLM) on the device, to analyze web pages and determine if they seem suspicious or scam-related.
This development was spotted by Leopeva64 on X, who regularly highlights web browser features which are being tested. What they actually discovered was a flag that can be enabled called ‘Client Side Detection Brand and Intent for Scam Detection,’ which is present in the latest version of Chrome’s experimental browser, Canary.
It looks like Chrome will also use AI to detect scams, there’s a new flag in Canary that suggests this:https://t.co/uQb84zbqeW pic.twitter.com/7CyK5y4gtODecember 19, 2024
The new flag leverages an on-device LLM to investigate the content of any given web page and figure out what it’s trying to do, and whether that content falls in line with the website’s supposed purpose or brand.
This is explained in the flag’s description, which reads: “Enables on-device LLM (large language model) output on pages to inquire for brand and intent of the page.”
On device is key to privacy
One key detail about this process is that it uses an on-device LLM, which means that the analysis of web pages happens on your device (as opposed to in the cloud somewhere, which would involve sending your browsing data to a third-party). In short, this means your data will stay private.
To try this feature out, you would have to install the latest Google Chrome Canary release, which is not something I’d generally recommend, unless you’re really keen (if so, you can follow Neowin’s advice on how to enable the new flag).
This is the latest in a series of AI-powered tools coming to Chrome, which also includes a ‘Store reviews’ feature that’s currently in testing. This capability uses AI to summarize reviews from platforms like Trustpilot or ScamAdvisor, helping users quickly check if an online store is reliable.
As ever, we don’t know if features in testing will make it through to release, but it’s likely these will – Google is keen on building out AI powers for its browser, and I expect we’ll see this scam warning system rolled out before too long. Unless the Google Chrome team finds some good reasons to go back to the drawing board.
Even though Chrome is the dominant web browser by a long way, Google shouldn’t rest on its laurels, and I think it’s very savvy of the company to keep improving its browser to stay in pole position. And with scammers and hostile actors now having AI-powered tools at their disposal, it’s good to see Google (hopefully) bringing LLMs in to help defend Chrome users from the unwanted attention of these nefarious types.
