JPMorgan Chase, a US-based bank, will not allow external artificial intelligence (AI) agents to access customer accounts and act on customers’ behalf. Jamie Dimon, CEO of JPMorgan Chase, announced the position during the bank’s third-quarter earnings call, citing risks to clients and the company, including the possibility of rogue agents operating in large numbers. A company spokesperson later confirmed that the bank currently blocks external AI agents from operating on behalf of retail customers. The decision comes as technology companies, including Meta, release autonomous assistants that can carry out financial tasks for consumers.
Read-only access through Open Banking providers
According to the bank, third parties may not access accounts by obtaining customers’ banking credentials and logging in on their behalf. JPMorgan Chase said it supports customer-authorised data sharing through Open Banking providers and is committed to enabling secure agentic connections. In practice, AI agents can still read some customer data, such as account balances, when Plaid or another connector holds the user’s credentials. However, they cannot send money or pay bills, and the bank has not set a timeline for safeguards that would allow direct account interaction.
The position recalls an earlier dispute with fintech companies over customer data, when JPMorgan Chase moved to charge data aggregators, including Plaid, for access. In both cases, the bank’s security concerns have been set against technology firms’ demands for account access.
Internal AI tools and liability considerations
The CEO said JPMorgan Chase uses AI agents internally and plans to deploy them in customer-facing applications, describing its approach as being about deploying the technology safely rather than opposing it. In recent weeks, the bank has started a targeted pilot of Smart Cash, an internal tool designed to forecast a user’s cash flow and move funds to maximise yield, according to its CFO. The bank also said that customers who authorise outside agents should understand the risks, including a potential transfer of liability.
Personal AI agents have become popular with mainstream users in recent months as they take on functions such as cancelling subscriptions or using credit cards on a user’s behalf. Instinct, a text-based digital concierge, launched on an invitation-only basis earlier in 2026, and its US-based developer, Spear Street Technology, reached a valuation of USD 10 billion in September 2026. That same month, bank stocks came under pressure after Meta’s Muse agent topped app download charts, amid concerns that consumers could use agents to move funds from low-yielding accounts to higher-yielding ones. OpenAI released its own agent, Dots, a few weeks later. Apollo’s chief economist was among those who flagged the risk to low-cost bank deposits, although that risk would be limited if agents cannot operate freely within customer accounts.
