Major U.S. financial institutions are scrambling to block artificial intelligence (AI) agents from logging into customer accounts and moving money.
Citigroup Inc. and Bank of America Corp. are the latest banking giants to formally restrict autonomous software from conducting direct transactions, joining Charles Schwab Corp. and JPMorgan Chase & Co. in establishing strict digital boundaries.
The defensive posture comes as financial executives raise alarm over security vulnerabilities, legal ambiguities, and systemic risks to traditional banking models.
The crackdown targets a new generation of viral personal AI assistants such as Meta Platforms Inc.’s Muse designed to handle daily tasks on behalf of users. Bank leaders argue that handing account credentials over to external algorithms exposes consumers and institutions to untenable risks.
“We feel very strongly that there should be a human involved in authorizing before money moves out of your account,” Bank of America CEO Brian Moynihan said in an interview last week, noting that unresolved liability questions make unauthorized automated transfers impossible to manage.
Echoing these concerns, Charles Schwab CEO Rick Wurster confirmed his firm actively blocks unauthorized agents ten to twenty times per week. “We can’t control it, the client can’t control it, something could go wrong and who’s liable?” Wurster cautioned, emphasizing that Schwab will not allow clients to share passwords with third-party bots.
Beyond immediate security concerns lies an existential commercial threat to commercial banking.
Automated AI agents are increasingly programmed to maximize consumer returns by routinely sweeping idle household cash out of low-yield deposit accounts into higher-yielding market alternatives.
Approximately $1.6 trillion in non-interest-bearing deposits held across JPMorgan Chase, Bank of America, and Wells Fargo % Co. could be jeopardized by automated reallocations, according to Apollo Chief Economist Torsten Slok.
The low-cost funding base forms the foundation for profitable commercial lending; forcing banks to compete with automated yield-seeking tools could saddle the sector with an estimated $47 billion in additional annual interest costs.
Rather than banning AI entirely, institutions are establishing gated, view-only frameworks. Schwab and JPMorgan are pursuing secure connections such as Model Context Protocol (MCP) servers that permit AI tools to view balances and conduct financial research without granting power of attorney to transfer funds or pay bills. Banks are also steering consumers toward proprietary, in-house tools, such as Schwab’s newly announced AI assistant, Charley.
The banking sector’s coordinated pushback mirrors recent actions by retail powerhouse Amazon.com Inc., which recently implemented technical blocks against Meta’s Muse and issued enforcement demands against AI search engine Perplexity AI’s Comet browser integration.
While federal regulators have largely supported the guardrails set by banks, officials acknowledge that comprehensive legislation will ultimately be required to establish clear authorization standards and clarify financial liability when autonomous software acts on a consumer’s behalf.
