Copyright © 2025 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{178562,
        author = {Aadrika Varshney and Lakshya Raj Singh Rathore and Antriksh Sharma and Saksham Shakya and Shubham Kumar},
        title = {Socio.io- A Web Based Extension},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {12},
        pages = {3938-3942},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=178562},
        abstract = {The developing use of unprotected sensitive textual data alongside vulgar content and gambling promotions and hate speech and Personally Identifiable Information (PII) together with AI-generated deepfakes constitutes a couple of threats to online integrity and privacy. The present detection systems do not deliver an integrated solution that identifies fake media alongside dangerous text content simultaneously during real-time communication. This paper introduces Socio.io, a multimodal AI device designed to address those demanding situations thru 3 middle innovations: (1) a Hybrid Deepfake Detection mechanism combining Vision Transformers (ViT) and Local Feature Analysis (LFA), achieving 96.4% accuracy at the Celeb-DF dataset; (2) Context-Aware Text Encryption that utilizes a modified BERT-LSTM model alongside Fernet-256 cryptography to stumble on and defend touchy text sorts; and (3) Zero-Knowledge Compliance that allows redaction of personal data without server-aspect garage, reducing privateness publicity with the aid of 72% compared to standard moderation APIs. The Socio.io platform executes coverage modifications by means of google cloud dealers and embraces Federated Learning protocols that ensure GDPR and CCPA compliance for neighbor content policies. The text classification performance reached accuracy rates of 89.7% when the machine processed social media data at a speed average of 93ms thus demonstrating higher system throughput compared to OpenAI moderation endpoint by 33%. The system brings scalability and efficiency to defeat evolving AI threats within online communications through its architecture-based deployment methods of browser extension or API gateway.},
        keywords = {Deepfake Text and Image detection and Encryption of personal data and Hate speech Encryption and PII Detection in NLP constitute.},
        month = {May},
        }