Recently, JD.com officially launched the Agent Autonomous Payment Protocol (A2P2), the first systematic protocol in China designed for intelligent agent autonomous payment scenarios. This marks the transition of AI payments from conceptual exploration to a standardized framework.
The protocol categorizes the autonomous payment capabilities of intelligent agents into six levels, from L0 to L5. L0 represents full user confirmation for each transaction, while L5 refers to the intelligent agent's complete autonomy in making and executing payment decisions. JD.com focuses particularly on the intermediate levels L3 and L4: L3 allows agents to initiate payment requests within a single task, but the system must decide whether to approve it based on user-defined boundaries; L4 expands authorization further, allowing the agent to independently complete payments as long as the amount, scenario, and user rules meet pre-set conditions.
In terms of security mechanisms, the A2P2 protocol introduces an "ARI (Agent Runtime Identity)" mechanism, which synchronously binds the real user, agent identity, and runtime environment information at the moment of payment, and performs real-time verification to ensure traceable payment responsibility, unique execution entities, and trusted runtime environments, thereby reducing the risk of identity forgery and malicious calls from the source.
At the same time, the protocol features a "funds carrier" isolation structure, completely separating the agent's operations from the user's main account. The system assigns the agent a separate restricted account with hard constraints on the maximum amount, usage scenarios, validity period, and payees. Even if the agent is maliciously controlled, it cannot exceed its permissions to access the main account. Users can view the fund usage of each agent in real time and revoke authorization at any time.
In terms of settlement and audit after the transaction, A2P2 binds payment results with task delegation vouchers, ARI identity information, execution decisions, and tokens to form a complete evidence chain, and introduces a "evidence chain" mechanism to uniformly record key facts, solving the problem of dispersed payment information and incomplete audits in traditional systems, thus achieving verifiable, traceable, and auditable AI transactions.
Industry experts generally believe that the release of the A2P2 protocol signifies that intelligent agent payments are moving from product-level capabilities toward infrastructure-level standards, and may become one of the key underlying protocols for AI commercialization in the future.
