Google Confirms Android Attack Warning — Powered By AI

Forbes

08-05-2025

When it comes to the attention of cybercriminals, you can bet that Google holds it. Whether it's the latest confirmed attack targeting Gmail users, password security threats, or zero-day vulnerabilities, Google is right there in tight focus. Of course, Google is constantly shining on the flip side of the cybersecurity coin as well, updating software as it fixes vulnerabilities, making changes behind the scenes in response to novel attack methodologies, and suspending malicious accounts as they are detected. Now, in a newly published report, Google has confirmed it is bringing new AI-powered security warnings to help defend against Android attacks.
Large language models are not new to Google, nor is using them to help protect users from scammers. A newly published report authored by Jasika Bawa, Google's Chrome group product manager, and Phiroze Parakh, a senior director of engineering with Google search, brings us right up to date, however. If you have ever wondered just how Google uses AI in the fight against search scams and attackers targeting Android and Chrome, this is likely the report you have been waiting for.
When it comes to search, for example, Google said that it is now blocking 'hundreds of millions' of scam results every single day, and AI protections have meant that it can now catch the malicious pages with a success rate that has increased by 20 times compared to before the AI was implemented. This, the report said, is because these large language models mean Google is able to 'analyze vast quantities of text on the web, identify coordinated scam campaigns and detect emerging threats,' to keep one step ahead of the attackers. Bawa and Parakh pointed to airline customer services as being an example of this in action, claiming an 80% reduction in scam activity when it comes to being tricked into calling a malicious telephone number.
Gemini Nano on desktop devices, in conjunction with the Enhanced Protection mode of Chrome's Safe Browsing feature, provides an on-device approach to scam defense. 'We're already using this new AI-powered approach to protect users from remote tech support scams,' the report said, continuing that Google hopes to expand the protection to Android users in the future. However, no date was given for when this might be.
There is some good news, though, for those readers who don't like playing the waiting game. Google has said that it is launching new AI-powered notification warnings for those using Chrome on Android devices. 'When Chrome's on-device machine learning model flags a notification,' Google confirmed, 'you'll receive a warning with the option to either unsubscribe or view the content that was blocked.' The idea being that this will help to protect against malicious and misleading notifications, and if it gets it wrong, hey, these things happen, you will still have the option of continuing to receive notifications from the site in question. Although there will always be privacy concerns when using AI of any kind, justifiable or not, there's no denying that the Android attack protection that will derive from the simple act of filtering out more nefarious notifications is most welcome.