MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641076882 A) filed by Abburu Sai Shivaprakash on June 22, 2026, for Tooltrust: A Deterministic Policy Enforcement Architecture For Secure Agentic Ai Systems Using Taint-Aware Authorization And Risk-Adaptive Access Control.
Inventor includes Abburu Sai Shivaprakash.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: The present invention relates to AgentPEP (Agent Policy Enforcement Point), a deterministic runtime authorization framework designed for securing autonomous and multi-agent Artificial Intelligence (AI) systems in enterprise environments. The system intercepts every tool invocation, API request, resource access operation, and inter-agent delegation at the execution boundary before actions are performed. The framework integrates a Universal Intercept API, a hybrid Role-Based Access Control (RBAC) and Risk- Adaptive Access Control (RAdAC) Policy Engine, a Taint-Aware Data Flow Tracker implementing session-based taint propagation with three trust levels (TRUSTED, UNTRUSTED, QUARANTINE), a Confused-Deputy Detection Engine for multi-agent delegation-chain validation, a Risk-Adaptive Authorization Engine computing composite risk scores across five weighted dimensions, and a Human-in-the-Loop Approval Module for selective escalation of high-risk actions. All authorization decisions are recorded in a tamper-resistant Audit Log Repository supporting GDPR, ISO 27001, SOC 2, DPDPA, and CERT-In compliance. AgentPEP implements deterministic security policies through verifiable software logic at the execution boundary, independent of model behavior. This approach enhances enterprise AI security by preventing unauthorized tool execution, indirect prompt injection, privilege escalation, and confused-deputy attacks, supporting trustworthy and governance-compliant AI deployments at scale. FIG.1
Disclaimer: Curated by HT Syndication.
