9/25/2026 Rudy San Miguel
Luyi Xing, CS associate professor at the Siebel School of Computing and Data Science in The Grainger College of Engineering, received a $1.5M National Science Foundation (NSF) Safe-OSE grant to advance security for AI Agent Ecosystem. Xing and his team are building on the Natural Language Interaction Protocol (NLIP) for communication with and between AI agents, for which Xing has held a leadership role with Ecma International and which will soon become an official ISO standard.
Written by Rudy San Miguel
Luyi Xing, CS associate professor at the Siebel School of Computing and Data Science in The Grainger College of Engineering, received a $1.5M National Science Foundation (NSF) Safe-OSE grant to advance security for AI Agent Ecosystems. The research being done by Xing, lead principal investigator (PI), along with his co-PIs, Siebel School CS associate professors Gang Wang and Xiaojing Liao, and Elisa Bertino, CS professor at Purdue University and Association for Computing Machinery (ACM) president, will soon be an official standard.
Xing's research has been in collaboration with Ecma International, a Swiss-based global association for standardizing industry information and communications systems whose standards include ECMAScript, the standardized foundation of JavaScript and ECMA-404, the JSON data interchange syntax. In June 2026, University of Illinois Urbana-Champaign was approved as a not-for-profit member, expanding Illinois’ role in developing international technology standards. Membership enables Illinois researchers to participate directly in the development and leadership of international technology standards across computing and information technologies. Xing will serve as the Illinois system’s principal representative to Ecma.
Ecma’s Technical Committee (TC56), for which Xing holds a leadership position as secretary, develops the Natural Language Interaction Protocol (NLIP) for communication with and between AI agents. NLIP standards, published by Ecma in December 2025, defines an open protocol for human-to-agent and agent-to-agent interaction across organizational and technological boundaries.
“NLIP addresses a central challenge in AI: enabling agents developed by different organizations using different AI models, tools, frameworks and operating environments to communicate,” Xing said.
NLIP recently received an approval vote from its member countries at the International Organization for Standardization (ISO), paving the way for it to soon become an official ISO standard.
“AI agents will increasingly need to work across organizations, AI models, tools and platforms,” Xing said. “Open standards such as NLIP can provide the common communication layer needed for interoperable and secure agent ecosystems. At the agent-ecosystem level, NLIP is designed to play a role analogous to HTTP/HTTPS on the Web: a common interaction layer that allows independently implemented endpoints to communicate through standardized messages.”
Xing and his team built on this standards work, leading to the NSF award to enhance AI agent security and build a secure open-source ecosystem around NLIP. The NSF project will develop and evaluate security protection at the agent communication protocol layer, using standardized message envelopes and control fields as a unified enforcement point for validation, authorization and policy constraints. The unified enforcement point enables the project to effectively address a broad range of critical security risks affecting agentic systems and LLMs. Target threats include prompt injection, jail breaking, semantic manipulation, authorization weaknesses, session hijacking, cost-amplification misuse and memory poisoning. The project will also produce open-source security extensions, validation tools and conformance resources for the broader NLIP and AI agent ecosystem.
“With Illinois’ recent Ecma membership, NLIP’s international standardization trajectory, my role in Ecma TC56 and the NSF Safe-OSE award, we can have a real impact,” Xing says. “This positions the Siebel School and Grainger Engineering to help shape secure, interoperable standards for next-generation AI agent ecosystems.”
Grainger Engineering affiliations:
Luyi Xing is an Illinois Grainger professor of computer science in the Siebel School of Computing and Data Science and an affiliate in Electrical and Computer Engineering.
Gang Wang is an Illinois Grainger professor of computer science in the Siebel School of Computing and Data Science and an affiliate in Electrical and Computer Engineering and the School of Information Sciences.
Xiaojing Liao is an Illinois Grainger professor of computer science in the Siebel School of Computing and Data Science.