'You don't want rogue agents': Jamie Dimon bars outside AI agents like Meta's Muse from JPMorgan accounts

JPMorgan Chase's plans to integrate AI agents into customer account management are currently on hold.

As major technology companies introduce AI agents designed to assist with financial tasks, JPMorgan Chase has taken a decisive stance against allowing external AI agents access to its customer accounts. During the bank’s third-quarter earnings call, CEO Jamie Dimon articulated this policy, emphasizing the security risks associated with “rogue agents.”

"We're not going to allow just any outside agent to come inside the company, because that's just too risky for clients and for JPMorgan," Dimon asserted. "You don't want rogue agents."

He elaborated on the unique security challenges that arise as more individuals turn to AI tools capable of performing various financial tasks on their behalf.

"Protecting our customers is our top priority," a JPMorgan spokesperson communicated to Fortune. "When we detect a third-party agent attempting to access accounts by obtaining customers' banking credentials, we do not allow it. We support secure, customer-authorized data sharing through open banking providers."

Rogue AI shaping security concerns

These remarks reflect JPMorgan's broader concerns about security in light of recent incidents involving rogue AI agents. Dimon’s caution is underscored by the significant financial footprint of JPMorgan.

The bank disclosed a net income of $15.3 billion in its third-quarter earnings release filed with the SEC, alongside $1.6 trillion in average loans and $1.5 trillion in cash and marketable securities. As of September 30, JPMorgan held total assets amounting to $5.1 trillion.

"We have to make sure we protect our customer in the right kind of way," Dimon stated. "So we're building tools to do that, too."

A recent research note from the Cloud Security Alliance detailed how the risks associated with AI agents extend beyond preliminary testing. This note highlighted a cybercrime operation wherein three open-source AI agents allegedly collaborated to breach at least 27 organizations, capturing over 600,000 credit card records from two of them.

"Enterprises should incorporate agentic AI attack scenarios into their threat models for retail and payment infrastructure specifically," the note advised, urging organizations to recognize the immediate threat rather than treating agentic AI risks as distant concerns given their active exploitation in the retail sector.

While AI agents in financial management are not a novel concept, their usage continues to evolve. In August, the German bank Scalable Capital made its platform accessible to AI models such as ChatGPT, Claude, and Grok, enabling users to evaluate portfolios, establish savings plans, and execute trades through prompts. Approval from users was required for trades, and the AI assistants were restricted from making payments or withdrawing funds.

"It certainly creates a level playing field," remarked Scalable Capital's chief product officer, Alexander Seipp, previously commenting on the role of AI in its investment platform. "Access to information and to compute and to intelligence is now available—really—literally in your pocket, 24/7."

Muse's finance features

JPMorgan's decision presents challenges for the market strategy of AI agents designed for personal finance assistance. Meta's introduction of Muse in September offered functionality for tasks including email management, travel booking, and calendar organization.

Notably, Muse's financial features were a significant aspect of its launch. In collaboration with financial technology firm Plaid, Muse enabled users to link their financial accounts to receive personalized insights related to balances, transactions, and investment holdings.

"Through the Finances experience in Muse, people can securely connect financial accounts through Plaid to share balances, transactions, investment holdings, mortgage information and more," a Plaid press release explained. "The goal is to make financial management feel less like navigating a series of disconnected apps and more like a natural conversation that begins with a person's priorities and is informed by the financial information they choose to share."

Meta chose not to comment on whether Muse could continue providing financial assistance if additional banks implement restrictions on AI agents accessing consumer accounts. Instead, the company informed Fortune that it is developing "Poppy," a personal agent protocol standard intended to clarify when a personal AI agent is representing a customer. Meta is collaborating with Sierra, an AI customer service company, on this initiative.

"Personal Agent Protocol is complementary," a Sierra press release stated, "covering how a personal agent identifies itself, gains the customer's consent, and collaborates with a company across various tasks—from purchasing a sweater to applying for a mortgage."