Major US banks are tightening controls on third-party AI agents accessing customer accounts, raising a broader question about the future of autonomous payments: how can AI systems move money without compromising security or user control?
On October 22, Reuters reported that Major US banks – including JPMorgan Chase, Bank of America and Morgan Stanley – were blocking or restricting third-party AI agents from accessing customer accounts. Wall Street institutions moved to block these autonomous tools over growing security concerns, as some agents appeared to act without explicit user instructions.
Sponsored
While the Reuters report did not establish that these incidents caused widespread financial losses, cited sources indicate that the restrictions stem from a fundamental technical hurdle: banks struggle to reliably verify an AI agent’s authorization and control its actions.
In several instances, customers had voluntarily shared their banking credentials with AI tools to execute automated tasks.
The clampdown follows broader regulatory momentum. Prior to this latest development, Reuters had reported that supervisory authorities across the US, UK, and the European Union were actively assessing the cyber threats AI poses to banking infrastructure, alongside banks’ reliance on tech vendors and their operational resilience following potential incidents.
Tech Sell-Off Weighs on Markets as Bank Stocks Stay Resilient
Despite the severity of the AI access restrictions, Wall Street bank shares did not react uniformly, showing no signs of a widespread banking sector sell-off. Shares of JPMorgan Chase, Bank of America, and Wells Fargo all edged up by less than 1%, while Morgan Stanley dropped approximately 2.1%.
The broader US market pulled back on October 22, led primarily by technology and AI stocks. The S&P 500 fell 1.2%—marking its worst single-day drop since July—while the tech-heavy Nasdaq Composite dropped 2.6%. The market data does not establish that the banking restrictions contributed materially to the sell-off.
TradFi vs. Open Standards: The Race to Authorize AI Payments
Although legacy banks are restricting third-party AI agents over security concerns, the underlying financial infrastructure is racing to define how verified AI agents will be permitted to transact.
Both traditional payment networks and crypto developers build competing approaches to authorized machine-to-machine payments.
Earlier this week Visa, BILL, and Billtrust announced the first live pilot project enabling AI agents to initiate and settle authorized B2B account payments on supplier portals. The initiative aims to automate corporate workflows while embedding authentication and control mechanisms directly into the process.
Concurrently, open-source and crypto infrastructure providers are developing alternative approaches. The x402 protocol, stewarded by an independent foundation under the Linux Foundation, is designed to enable internet-native payments between applications and AI agents. It supports multiple payment methods, including stablecoins, and aims to make machine-to-machine transactions easier to initiate over the web.
These approaches are not necessarily direct competitors: traditional payment networks and open payment standards operate at different layers and may be complementary. Both, however, face a shared challenge as AI agents take on more financial tasks: making sure agents only spend money when authorized, keeping their actions within set limits, tracking what they do, and deciding who is responsible when things go wrong.
Discover DailyCoin’s popular crypto news today:
$1.62 Trillion German Banking Giant Launches XRP Trading
Google Cloud London Outage Knocks Out Revolut, Bank Services