SurakshaNet detects abusive messages on-device, protecting privacy while enabling fast reporting.

What is it about?

The work is about detecting abusive or harmful messages in real time without sending users’ private conversations to the cloud. The proposed system, SurakshaNet, is a lightweight Chrome extension designed for platforms such as WhatsApp Web. It can identify abusive content written in Hindi, English, and mixed Hindi-English language, and warn the user when a message appears harmful or critical. Its main contribution is that the detection happens locally inside the browser. This improves privacy because raw messages remain on the user’s device. The AI model was compressed from 471 MB to 113 MB, enabling fast browser-based detection with a median response time of about 28 milliseconds and very little loss in accuracy.

Featured Image

Why is it important?

It is important because online abuse can spread quickly and seriously affect users, particularly women, children, and other vulnerable groups. Many existing moderation tools send messages to cloud servers for analysis. That creates two problems: private conversations may leave the user’s device, and the warning may arrive with a delay. SurakshaNet addresses both issues by detecting abusive content directly inside the browser. Its importance comes from four practical benefits: Privacy: users’ raw messages remain on their own device. Speed: harmful messages can be detected almost immediately, with a median response time of about 28 milliseconds. Language support: it can recognise abuse in English, Hindi, and mixed Hindi-English conversations, which are common in Indian online interactions. Evidence preservation: serious incidents can be recorded securely with timestamps and tamper-resistant logs, making reporting easier. Overall, the work shows that it is possible to build a fast, privacy-friendly safety tool that can protect users without continuously sending their personal conversations to external servers.

Perspectives