JPMorgan Chase (NYSE: JPM), which remains the largest US bank by assets, has moved to block outside artificial intelligence agents from gaining direct operational access to customer bank accounts.
Chief Executive Jamie Dimon outlined the position during the bank’s third-quarter earnings discussion on October 13, 2026, emphasizing that unrestricted entry by external agents would expose both clients and the institution to unacceptable hazards.
Dimon stated that the bank would not permit just any external agent to operate inside its systems, describing the scenario as excessively risky for customers and for JPMorgan itself.
He specifically highlighted the danger of “rogue agents,” noting that such systems could number in the thousands and pose a broad threat.
The comments come amid the rapid emergence of autonomous personal assistants from technology companies, including Meta’s Muse, which launched in September 2026 and quickly climbed download charts, raising industry concerns about potential automated shifts of deposits from low-yield accounts to higher-yielding alternatives.
A bank spokesperson later confirmed to CNBC that JPMorgan already prevents external AI agents from acting on behalf of retail customers.
The institution does not allow third parties to obtain customers’ banking login credentials and sign in on their behalf, citing safety as the primary concern.
At the same time, the bank supports secure, customer-authorized data sharing through established open banking intermediaries and has expressed a commitment to enabling agent-based connections in a controlled manner once appropriate protections exist.
In practice, this means certain AI tools may retrieve limited account information, such as balances, on a read-only basis when a customer has previously authorized a connector service like Plaid to hold their credentials.
However, these external agents cannot initiate money transfers, pay bills, or perform other transactional actions that would make them fully functional for day-to-day financial management.
The spokesperson indicated that no specific timetable has been set for implementing the additional safeguards required to permit direct agent interaction with accounts.
Dimon noted that JPMorgan itself employs AI agents for internal purposes and intends to introduce them in customer-facing applications as well. Chief Financial Officer Jeremy Barnum reported that the bank recently launched a targeted pilot of its own Smart Cash tool.
This system is designed to project a customer’s cash-flow needs and automatically reallocate funds to optimize returns, effectively serving as an in-house alternative to third-party agents.
Dimon further advised that customers who choose to authorize outside agents should fully understand both the associated risks and the potential shift in liability that accompanies such arrangements.
The bank’s stance echoes earlier tensions with financial technology firms over customer data access, including disputes involving data aggregators.
Examples of the external agents drawing attention include Meta’s Muse and Instinct, a text-based digital concierge developed by Spear Street Technology that recently achieved a $10 billion valuation.
By drawing a clear boundary around direct account control, JPMorgan is prioritizing security and customer protection while continuing to develop its own agentic capabilities and exploring secure pathways for future integration.