How Ant Group’s ASL Protocol Bridges the Trust Gap in Multi‑Agent Collaboration

The IIFAA Internet Trusted Certification Alliance, together with Ant Group, China Academy of Information and Communications Technology, and more than twenty core industry partners, officially published the Agent Security Link (ASL) protocol, delivering a practical, verifiable security architecture for native AI agents.

As multi‑agent collaboration moves toward cross‑platform, complex scenarios, issues such as identity spoofing and command tampering become prominent trust holes. ASL is designed as an enterprise‑grade security component that can be layered on existing interconnection protocols like MCP and A2A, constructing an end‑to‑end protection system covering the entire agent interaction workflow.

Verifiable identity and runtime joint verification: Agent identity credentials are jointly validated with runtime environment state across domains, preventing identity forgery and environment tampering.

Session protection for agent collaboration: Mutual two‑way authentication establishes secure sessions, ensuring message integrity, ordered delivery, and resistance to replay attacks.

Controlled authorization boundaries and multi‑level delegation: A strict permission‑shrinking mechanism guarantees that delegated sub‑agents only receive reduced privileges, preventing cumulative over‑reach.

Semantic consistency of intent across the full chain: The original user intent remains immutable as it traverses agents; each execution generates a verifiable receipt, providing an auditable trail.

ASL’s value is demonstrated in high‑frequency AI services such as parking fee payment, utility bill settlement, and shared‑mobility on smartphones and AI glasses. For example, a user wearing an AI glass can say “pay my parking fee,” and the agent will automatically handle identity confirmation, intent transmission, authorization, and receipt generation within the trusted scope. In payment scenarios, ASL works alongside the ACT Intelligent Agent Commercial Trust protocol—ASL secures inter‑agent connectivity and authorization, while ACT builds the commercial transaction trust foundation.

The protocol ecosystem is intended to evolve continuously. Built on open and co‑creation principles, ASL invites industry partners to keep pace with technological trends, deepen scenario innovation, and drive protocol iteration, making trustworthy interconnection a foundational infrastructure for the intelligent world.

The IIFAA Internet Trusted Certification Alliance was launched in 2015 by the China Academy of Information and Communications Technology, Alibaba, Huawei, ZTE, Ant Group, Ping An Technology, and other institutions, now comprising over 300 member organizations.