JPMorgan moves to limit outside AI agents and shifts a deposit token to Ethereum, aiming to tighten control over bank-grade systems and payments

JPMorgan Chase is reportedly taking steps to limit the ability of “outside” artificial intelligence agents to operate within its systems, while also moving a JPMorgan-issued deposit token onto Ethereum. The moves, described in a recent market report, are framed as protections for the bank’s operational control and the reliability of token-based settlement.

According to the report, JPMorgan is “blocking outside AI agents,” a phrasing that suggests the bank is tightening permissions around autonomous or semi-autonomous software tools that may be run by third parties. In practice, banks typically treat AI agents differently from traditional software because agents can initiate actions across multiple systems. Limiting them can reduce the chance of unauthorized changes, unintended data exposure, or automated workflows that violate internal policy.

The same report says JPMorgan is taking its deposit token to Ethereum. A deposit token is essentially a digital representation of a deposit claim, intended to make it easier to move value and synchronize payment activity with other digital systems. Ethereum, a widely used public blockchain platform, can enable programmable transactions, but it also introduces a different operating model than JPMorgan’s own or permissioned networks.

The market report also links the two technology directions to the idea of “what both moves protect,” positioning the changes as safeguards rather than standalone tech experiments. The logic is straightforward: autonomous AI tools increase the need for access controls and auditability, while public-chain deployment increases the need for robust settlement, key management, and clear operational boundaries.

JPMorgan has been involved in payments and blockchain experimentation for years, and industry observers have generally associated tokenized deposits with potential benefits such as faster interbank settlement and improved reconciliation. Ethereum, in particular, is used by a large ecosystem of developers and applications, which can speed integration. At the same time, public networks can raise concerns around transaction finality timing, transparency of transaction metadata, and the governance of the software stack used to issue and redeem tokenized claims.

The reported steps come with notable disclosure limits. The market post, as presented in the available packet, does not provide technical specifications such as which “outside” AI agents are being restricted, what enforcement mechanism is being used, whether the limitation applies to specific vendors only, or how the bank tests the compliance of AI-driven workflows. It also does not detail the exact token contract architecture on Ethereum, how JPMorgan handles mint and redemption controls, or the operational process for reconciling token movements with traditional ledger records.

From a sector perspective, the combination of AI governance and token settlement reflects how financial institutions are adapting to two parallel pressures. First is the accelerated deployment of AI tools in enterprise environments, where the key issue is controlling who or what can cause system changes. Second is the search for usable rails for digital value transfer, where the key issue is making token settlement resilient enough for bank-grade operations.

What to watch next is whether JPMorgan provides more concrete, auditable descriptions of the permissions model around AI agents and the risk controls around its deposit-token operations on Ethereum. Investors and clients will likely look for indicators such as detailed governance frameworks, clearer operational procedures, and any measurable changes in performance or reliability once the Ethereum deployment is fully integrated.