PBOC Deputy Governor Lu Lei told the 15th China Payment and Clearing Forum that agent-based payment cannot blur the responsibility boundaries between consumers, payment institutions, and algorithmic systems, according to a ChainCatcher report citing Mobile Payment Network.
Lu, a member of the PBOC Party Committee, said payment is essentially a transfer of fund ownership. That makes it necessary for transactions to produce predictable outcomes, assignable responsibility, and traceable trails. Large models and autonomous agents have features such as random outputs and opaque logic. If trading decisions are blindly or excessively delegated to them, the trust foundation of fund transfers could be damaged.
Rules built around human decision-makers need updating
Current payment governance rules and dispute resolution mechanisms are built around “humans as the final decision-makers in transactions.” The new model where agents automatically initiate or assist transactions can easily blur responsibility lines, Lu said. Existing governance rules therefore need to be optimized.
Standards and the self-regulatory convention
On the lack of compatibility between protocols and standards in the agent payment field, Lu stressed that the protocol contest is essentially a contest over business rules and technical standards, and also a contest for dominance in the artificial intelligence era.
The PBOC has been strengthening follow-up research on tech innovation, especially agent payments. It has guided the Payment and Clearing Association to formulate and release the Self-Regulatory Convention on Agent Payment Applications after broad consultation. Going forward, the central bank will keep working on coordinating protocols and standards and on innovation risk governance.
Three expectations for market institutions
Lu put forward three expectations:
- Actively respond to and implement the industry self-regulatory convention, with payment security and risk prevention as the baseline and consumer protection as the goal;
- Keep tracking cutting-edge technology trends such as large models and agents at home and abroad, and build up technical reserves and application capabilities;
- Put rules and standards first, improve coordination and compatibility across different protocols and standards, and support regulators in building the underlying protocols and technical standard system for agent payments.