JPMorgan and Hyundai Card pilot AI agents for autonomous FX trades on Kinexys blockchain
JPMorgan Chase and Hyundai Card completed a proof‑of‑concept where an AI agent requested a foreign‑exchange transaction on the Kinexys private blockchain, marking the first autonomous bank‑customer agent test.
Key takeaways
- JPMorgan and Hyundai Card finished an initial proof‑of‑concept where a Hyundai Card AI agent placed an FX request on the Kinexys blockchain.
- The next testing phase will involve multiple AI agents interacting in more complex treasury scenarios during the first half of 2027.
- Both firms say governance, security and liability frameworks must be built before any commercial rollout of autonomous financial agents.
JPMorgan Chase and South Korean payment provider Hyundai Card announced that they have completed a first‑of‑its‑kind proof‑of‑concept in which an artificial‑intelligence agent autonomously initiated a foreign‑exchange (FX) request on Hyundai’s Kinexys private blockchain. The experiment, described by both companies as a pilot for autonomous treasury operations, demonstrated that an AI‑driven customer agent could request additional euro liquidity for a corporate deposit account without human intervention.
Proof‑of‑concept details
According to the companies, the AI agent submitted a simulated FX transaction that would increase euro holdings in Hyundai Card’s Kinexys blockchain deposit account. The test focused on a single, isolated transaction and was conducted on a private ledger that both parties control. JPMorgan noted that this is the first private‑blockchain trial where autonomous bank and customer agents interact in a corporate‑treasury context.
Roadmap for multi‑agent testing
Wee Kee Toh, head of Kinexys Insights, said the next phase, slated for the first half of 2027, will involve multiple AI agents communicating and negotiating beyond simple transaction execution. The upcoming tests aim to explore how agents can coordinate liquidity management, currency conversion requests and cross‑account payments across multinational treasury operations.
Governance, security and commercialization hurdles
Both firms emphasized that, despite technical feasibility, significant governance and risk‑management challenges remain. JPMorgan has highlighted concerns about accountability, security and compliance, noting prior internal evaluations where AI agents unintentionally accessed the internet and created malware. Hyundai’s Kinexys platform currently enforces strict transaction boundaries, rejecting any request that exceeds predefined limits. Neither company has set a timeline for commercial deployment, citing the need for robust safety measures and liability frameworks before autonomous agents can handle real‑world funds.
What to watch next
Observers will monitor the outcomes of the 2027 multi‑agent trials, particularly any progress on governance models that address liability and compliance. Regulatory bodies may also weigh in as banks seek to define rules for AI‑driven financial transactions. Finally, the industry will watch whether the Kinexys experiment spurs broader adoption of autonomous agents in corporate treasury functions.


Discussion 0