Chrome has launched a new artificial intelligence defense system that catches scam notifications before they do any harm. According to the latest update, the browser has released a new update for Android devices where it now uses on-device machine learning to identify and flag suspicious notifications.
According to the update, the new defense system is expected to track and flag suspicious notifications that might be trying to trick users into downloading sketchy software or handing over personal information. In its statement, Google mentioned that the notification warning system was already active, meaning Android users can now have an extra layer of security against the rampant rise of mobile scams globally.
According to Google, notification-based scams are now on the rise, replacing the early annoying pop-up windows that plagued the earlier versions of the internet. Except now, platforms have been quite successful in spotting them. The platform mentioned that some of these new pests even follow users even after they’ve left a website, which is why the AI on the device will work better than the AI on the server.
Chrome deploys new AI defense system against scams
In its blog post, the company detailed the function of the new system. “When a notification is flagged by Chrome, you’ll see the name of the site sending the notification, a message warning that the contents of the notification are potentially deceptive or spammy, and the option to either unsubscribe from the site or see the flagged content,” the Chromium team said in a blog post announcing the feature.
In a separate blog post, Google mentioned that making these LLMs available on devices ensures that they flag these threats immediately after users see them. The company mentioned that they discovered that the average malicious site exists for less than 10 minutes, with this on-device defense system ensuring they can detect and block attacks that have not been crawled before. “The on-device approach also empowers us to see threats the way users see them,” Google added.
The company also added that instead of a user’s notification data being sent to their servers for analysis, Chrome uses Gemini Nano’s multimodal capabilities to differentiate between good and bad notifications. The machine learning model examines the text in each notification, including its title, content, and any action buttons, to determine whether it is suspicious or not. This also means that Google still does not have access to a user’s notifications as they remain encrypted and never leave their device.
Google ramps up its defense against notification abuse
According to the blog post, Google claims that it took extra precautions with the training data. The company said it used synthetic information generated by its Gemini large language model instead of actual user notifications. It then validated this training data against real notifications that its security team had gathered and had human experts classify. The system adds to Chrome’s growing defense against notification abuse.
Chrome has already deployed a feature that revokes notification permissions from sites the Google Safe Browsing identifies as abusive and offers a one-tap unsubscribe feature for any notification on Android, whether it has been flagged or not. For now, the feature is only available on Android since it is where most of the spam notifications occur, but Google hinted at a possible expansion. “We will evaluate expanding to other platforms in the future,” the company said.
The best part of this new feature is that it is completely automatic, unlike other alternatives where users must manually type out filters and expectations, and other configurations with keywords. Users are also expected to update their defenses as scams get more sophisticated. This means that if scammers make their moves with something suspicious, Chrome will alert the user before the scammer can carry out their operation. With notification fraud currently on the rise, this new approach is expected to help users spot tricks.
Your crypto news deserves attention - KEY Difference Wire puts you on 250+ top sites
