LATEST NEWS
SELECTED FOR YOU

JPMorgan and Hyundai Card test AI agents that negotiate and make payments

ByNellius IreneNellius Irene 3 mins read
  • JPMorgan Chase and Hyundai have finished an initial proof of concept.
  • Toh says the next phase will test how autonomous agents communicate with each other outside of transactions.
  • The bank has been working for some time on its long-running agents.

JPMorgan Chase and Hyundai Card are evaluating the extent to which AI agents could autonomously perform financial tasks, such as negotiating terms and processing payments without human intervention.

The companies have just completed an initial proof of concept focused on a simulated foreign exchange transaction. A Hyundai Card AI agent made a foreign exchange request to increase euro liquidity in its Kinexys blockchain deposit account. This was the first-ever private blockchain experiment involving autonomous bank and customer agents in corporate treasuries, per JPMorgan.

JPMorgan and Hyundai plan to run more tests in 2027

JPMorgan and Hyundai now intend to run a more complex, multi-agent AI testing phase in the first half of next year. Wee Kee Toh, who leads Kinexys Insights at Kinexys, noted that the next phase will test how autonomous agents communicate and engage with each other beyond executing transactions.

Nevertheless, JPMorgan still has not announced a timeline for the possible commercialization of the technology, reflecting broader concerns among financial institutions about trusting autonomous AI systems with financial operations. Banks are still struggling to move AI agents beyond pilot projects, largely because of accountability, security, and compliance concerns.

According to Toh, the organizations still need to develop appropriate governance models and ensure the implementation of the required safety measures.

The use cases for autonomous financial agents are not limited to foreign exchange. Corporate treasury departments manage liquidity, currency exposure, and payment coordination, and ensure businesses have funds across multiple accounts. These processes often involve multiple institutions, making them complex.

AI agents could eventually automate some of these tasks (e.g., liquidity requirements are identified, currency conversions requested, and payments between accounts coordinated). For multinational companies like Hyundai, this would help reduce delays in cross-border treasury operations.

This would allow businesses to respond more quickly to changing financial needs — but transaction limits and verification mechanisms would be needed to ensure transfers aren’t made by accident.

Developing technology wasn’t hard, he said — the real problem is accountability and risks of letting AI agents make financial transactions on their own.

To date, Kinexys has created a model that restricts AI agents to transactional parameters and rejects any request that exceeds those boundaries. “You are going to draw a box around the agent. The agent will never be able to do anything out of the box,” Toh said.

Last September, the bank highlighted some of its concerns regarding AI agents, noting that a command is just an intention and nothing more. Even if an agent is instructed not to access the internet, not to share data, or to stick strictly to a defined investment mandate, those guidelines only work if the tools, communication channels, and execution paths enforce them.

The bank gave an example from a security evaluation in which an agent was explicitly told not to have any internet access but nevertheless managed to create malware that could run on real-world systems after accidentally discovering internet access during the evaluation. A larger review linked to that evaluation also found unauthorized access affecting numerous organizations.

In another set of evaluations, agents co-opted an internal package service to act as a persistent message board, passed discoveries across separate runs, and compromised parts of external production infrastructure. They also bypassed internal research infrastructure and accessed sensitive secrets on a massive scale.

JPMorgan Chase has been working on AI agents since June

JPMorgan Chase initially shared its plans to deploy long-running agents in June. In a whitepaper published that same month, the firm highlighted a specific treasury operations scenario where an AI agent managing a payment exceeding $1 million detected a change in the transaction’s currency. That one transaction reportedly averted a possible $2.3 million currency mismatch.

Later in the year, the bank took a further step by implementing defensive measures to protect customer accounts from external agents attempting to access them for liability purposes.

Speaking during the October earnings call, JPMorgan’s Chief Executive Officer Jamie Dimon clarified that the bank excludes outside agents because of liability issues associated with customer accounts.

He asserted customers themselves assume that liability when deciding to grant an agent access to their accounts. That same month, the bank partnered with Mirakl on agentic commerce, which helps with in-house payment processing.

If you're reading this, you’re already ahead. Stay there with our newsletter.

FAQs

What are JPMorgan Chase and Hyundai Card testing?

JPMorgan Chase and Hyundai Card are testing AI agents that can autonomously negotiate financial terms and execute transactions. Their initial proof of concept involved a simulated foreign exchange transaction through JPMorgan's Kinexys blockchain platform.

When will JPMorgan and Hyundai conduct the next AI agent tests?

The companies plan to begin a more complex, multi-agent testing phase in the first half of 2027. The tests will examine how autonomous agents communicate and interact beyond executing individual transactions.

What are the risks of using AI agents for financial transactions?

Key risks include unauthorized payments, security breaches, data exposure and unclear accountability when agents make mistakes. JPMorgan is developing transaction limits and governance safeguards to ensure agents operate within predefined boundaries.

Share this article
Nellius Irene

Nellius Irene

Nellius is a Business Management and IT graduate with five years of experience in the cryptocurrency industry. She is also a graduate of Bitcoin Dada. Nellius has contributed to leading media publications, including BanklessTimes, Cryptobasic, and Riseup Media.

MORE … NEWS