Ant International has begun working with Mastercard and Visa on a “know your agent” framework intended to help payment networks identify trusted AI agents across digital wallets, card systems and online marketplaces.
The initiative was highlighted by Zhuoqun Bian, president of Ant Digital Technologies, at the Fortune Leaders Forum in Macau on September 8. The companies are collaborating through BuildFin.ai, an industry platform convened by the Monetary Authority of Singapore.
Bian said financial institutions would need to establish which AI agents were acting in a transaction, who owned them and whether they had been authorised to act.
“From the financial institutions’ perspective, when we initiate a transaction, we have to do the KYC [know your customer]. In the agent economy, you need to know your agents. Who’s the agent? Who does it belong to? Who authorized it?” she said.
The development comes as AI agents are expected to play a larger role in consumer payments. A January report from McKinsey estimated that agents could orchestrate as much as $5 trillion in global consumer spending by 2030.
However, the payment infrastructure used by agents was built for people, Bian said, meaning existing systems would need to be rebuilt or enhanced to support them.
The International Monetary Fund has also identified a potential tension between the way AI agents operate and the requirements of financial systems. In an April note, it said agents were probabilistic and adaptive, so the same prompt could produce different answers, while payment systems needed consistent results.
Card networks and real-time gross settlement systems depend on predictable rules, legal certainty and clear accountability, the IMF note said.
For banks, the principal challenge may lie less in the underlying technology than in the controls surrounding its use. Benson Wong, managing director and head of digital at JPMorgan Private Bank, said the risks were linked to operating models, processes, compliance and safeguards.
He said an incorrect response to an information request might result only in an embarrassing moment, but errors in automated workflows could have much wider consequences because of the agent’s greater autonomy.
“You need to know your agents much more than you know your customers,” Bian said.
