Google Chrome’s new on-device machine learning model can block 2.5x more potential phishing attacks and potentially malicious sites


Photo by Deepanker Verma from
This Article is written as a summay by Marktechpost Staff based on the Google article 'Building a more helpful browser with machine learning'. All Credit For This Research Goes To The Researchers of This Project. Other references post 1, post 2.

Please Don't Forget To Join Our ML Subreddit

The goal of machine learning (ML) has always been to develop more valuable products for finding answers to business-related questions. Google has been using machine learning to boost the use of its products, especially Chrome, since the dawn of time. Web images have been made more accessible to visually impaired people, and real-time captions for online movies have been generated for people in noisy or hearing-impaired situations. Google has made Chrome safer and more accessible with recent technological advancements while providing a more personalized browsing experience for its customers. Google values ​​the privacy of its customers above all else, which is why the latest enhancements are powered by on-device ML models, ensuring that user data remains private and never leaves their device. Every day, Chrome Safe Browsing protects billions of devices by displaying warnings when users try to visit risky sites or download dangerous files. Compared to the previous model, Google just announced a new machine learning model that can identify 2.5 times more potentially dangerous sites and phishing attempts, resulting in a safer and more secure web.

The importance of online notifications and how people interact with them was also highlighted. Page notifications allow users to receive updates from sites they visit regularly. Notification permission requests, on the other hand, can quickly become boring. Chrome predicts when permission prompts are unlikely to be given based on how the user has previously behaved with similar permission prompts and silences these unwanted prompts to allow consumers to browse the web with minimal disruption. Users can expect an ML model that produces these on-device predictions in the future version of Chrome. Google also developed Journeys to help people retrace their steps online by bringing together all the pages they’ve visited on a specific topic and making it easy to pick up where they left off. Research is also underway to make these websites available in the user’s language of choice. This will be accomplished by using a language identification model to determine the language of the page and whether it should be translated to reflect the user’s choices. This resulted in tens of millions of successful translations per day.


Google aims to create a truly and consistently useful browser. With the immense possibilities offered by machine learning, there are many untapped areas that academics want to explore further. Continuous research is underway to allow Chrome to become a browser designed specifically to meet all their demands. Future work will include applying machine learning to update the toolbar in real time highlighting the most convenient action at that time. Link sharing, voice search, and other customizations are among them.

Previous postChinese Researchers Introduce “FedPerGNN”: A Novel Federated Graph Neural Network (GNN) Framework for Efficient, Privacy-Preserving Personalization

Khushboo Gupta is an intern consultant at MarktechPost. She is currently pursuing her B.Tech from Indian Institute of Technology (IIT), Goa. She is passionate about the fields of machine learning, natural language processing and web development. She likes to learn more about the technical field by participating in several challenges.


About Author

Comments are closed.