PrivAware – An AI-Powered Privacy Policy Simplifier

The ever-growing online threat landscape means that current security systems that rely on fixed blocklists are not adequate in combating modern phishing attacks and shady privacy practices. PrivAware  addresses this by integrating real-time anti-phishing with privacy policy evaluation into an easy-to-understand Chrome extension.   Using transformer-based natural language processing (NLP) models with DistilBERT and the API system it enables, the product provides real-time, accurate, and interpretable risk scores for any web page. PrivAware  provides a single risk score that combines phishing, privacy policy and tracker information, allowing users to assess and mitigate cyber risks.   Evaluation results demonstrate that PrivAware has high precision, recall and accuracy, indicating it effectively identifies malicious links and potentially deceptive privacy statements. The system's integration with a browser extension makes it easy to use in real-time for non-technical users.   With future technological improvements in explainable AI, multi-language support and user preferences, PrivAware may become a user's personalised digital privacy coach. In conclusion, the system is a step in the direction of proactive, intelligent and user-friendly web security, bringing cybersecurity to real users.

●                 K. Vander/Schyff, G. Hoepman and S. Foster, Privacy policy analysis: A scoping review and Research agenda, Computers and security, vol. 146, p. 104065, February 2024. ●                 A. Adhikari, S. 19th Annual Int. Conf. on Privacy, Security & Trust (PST), 2022. ●                 Y. Shwartzshnaider, A. Balashankar, V. Patidar, T. Wies and L. Proceedings of the Association of Advancement of Artificial Intelligence (AAAI) annual meeting, Human Computation and Crowdsourcing (HCOMP), 2020. ●                 H.Cui, R.Trimananda, A.Markopoulou and S.Jordan, PoliGraph: Knowledge Graph-based Privacy Policy Analysis, arXiv:2210.06746, Oct. ●                 Privacy Parser: Deep Learning-based Privacy Parser, arXiv:2501.10319, Jan. ●                 "Summary of privacy policy based on Ensemble learning, in Proc. CODASPY, 2016. ●                 J. Saka, B. A. Ajayi and S. M. Marafa, Analysis of Privacy Policy of Corporations using Natural Language Processing and Deep Learning, Inter. J. Adv. Res. Computer. Eng. Technol., vol. 4, no. 12, 2023. ●                 The elaboration of the Websites Privacy Policies ( 2009-2019): Insight, Information, vol. ●                 Automating Privacy Policy Analysis with LLMs: Prompt engineering, Fine tuning and Explaining arXiv:2503.16516, Mar. ●                 J.Bhutto, Extraction and Analysis of Privacy Policies Automation B.A. Thesis, Univ. of Twente, 2021. ●                 P.  John and G.  Axenbeck, "An automated tool to measure the completeness of privacy policies," in Proc. Privacy in the Electronic Society, 2012. ●                 S. Sousa and R. Deep learning of privacy-preserving natural language processing: a review," Artif. Intell. Rev., May 2022.