MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202621065848 A) filed by Ajaj Khan; Rashmi Choudhary; and Adnan Khan on May 25, 2026, for Ai-Driven Adaptive Zero-Trust Cybersecurity System Using Federated Machine Learning.
Inventors include Ajaj Khan; Rashmi Choudhary; and Adnan Khan.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: The present invention discloses an AI-driven adaptive zero-trust cybersecurity system using federated machine learning. The system comprises distributed security agents that collect local telemetry and train local threat-detection models, a federated aggregation server that combines privacy-protected model updates into a global model using robust secure federated averaging without accessing raw data, an adaptive trust-scoring engine that continuously computes a time-varying trust score for each entity from behavioural, contextual and reputational signals, and a dynamic policy-enforcement point that translates the trust score into risk-proportionate access decisions for every transaction. The invention enables multiple organisations to collaboratively learn emerging threats while preserving data privacy and residency, tightly couples continuous detection with automated enforcement, and reduces false positives and attacker dwell time, thereby providing scalable, privacy-preserving collective defence across enterprise, cloud and Internet-of-Things environments.
Disclaimer: Curated by HT Syndication.
