• JPMorgan warns rogue AI agents could threaten banking security
• Jamie Dimon rejects unrestricted access for external autonomous agents
• Wells Fargo emphasizes disciplined risk management for third-party AI
• Goldman Sachs calls for rigorous testing and human accountability
• Citi reports AI accelerating product launches by 30% to 40%
• Citi cuts technology vendor spending by 30% year-over-year
• Morgan Stanley highlights productivity gains and customer trust
• Bank of America expands automation amid regulatory scrutiny

Wall Street's biggest banks are accelerating their adoption of agentic artificial intelligence while warning that autonomous systems could introduce significant security, governance and operational risks without appropriate safeguards.

JPMorgan Chase CEO Jamie Dimon told investors during the bank's third-quarter earnings call that the institution was embracing AI agents for employees and customers but remained concerned about the potential consequences of allowing autonomous systems unrestricted access to banking infrastructure.

Dimon said JPMorgan was not "against agents or for agents," emphasizing that the bank was balancing technological innovation with security considerations.

The possibility of rogue agents operating across financial systems represents a particular concern, he warned, especially as the technology becomes more widely deployed.

"We're not going to allow just any outside agent to come inside the company because that's just too risky," Dimon said.

JPMorgan is also seeking to avoid excessive dependence on individual AI providers by maintaining a model-agnostic strategy.

Its approach includes routing queries toward the most efficient and economical models rather than automatically using the most sophisticated technology available.

"You don't need the most expensive thing for every query you have," Dimon explained.

Other major US banks are confronting similar challenges as agentic AI moves beyond experimentation into operational deployment.

Wells Fargo CEO Charlie Scharf said the bank was enthusiastic about AI's commercial potential but was "taking a disciplined approach to managing risk."

The institution is evaluating third-party models and autonomous agents while assessing the potential vulnerabilities they could introduce for customers and internal operations.

At Goldman Sachs, CEO David Solomon acknowledged continuing uncertainty surrounding AI's development and deployment, arguing that financial institutions must establish effective controls without unnecessarily restricting innovation.

"The choice is not between halting progress and moving forward without guardrails," Solomon said.

He emphasized the importance of "rigorous testing, independent review, continuous monitoring and clear human accountability."

Despite these concerns, banks are reporting substantial productivity improvements and cost reductions.

Citigroup CEO Jane Fraser said AI had accelerated product launches by an average of 30% to 40%, while approximately 14,000 engineers were using the bank's Devin agentic coding platform.

Citi is also redesigning 100 manual processes using AI, alongside efforts to reduce dependence on external technology providers.

"We've reduced our tech vendor spend by 30% year-over-year as a consequence, and that's still got a long way to run," Fraser said.

Morgan Stanley CEO Ted Pick described AI's potential to improve efficiency, productivity and resilience as undeniable, although he stressed that customer confidence remained essential in financial advisory relationships.

The technology could undertake preliminary analysis and support financial decisions, he suggested, but trust between advisers and clients would remain fundamental.

Morgan Stanley CFO Sharon Yeshaya said AI was "fundamentally changing what our teams do," enabling employees to operate at greater scale while strengthening risk management.

Bank of America is similarly expanding automation and simplifying workflows, with executives indicating that productivity improvements are reducing the need to replace departing employees.

However, the bank's technological ambitions coincided with significant regulatory scrutiny. It was fined $1.21 billion for violations of anti-money laundering and sanctions requirements shortly before its earnings announcement.

The contrasting developments highlight a growing challenge for financial institutions: delivering measurable efficiencies through increasingly autonomous technology while maintaining regulatory compliance, operational resilience and accountability.

As adoption accelerates, the central question is becoming less about whether banks should deploy AI agents and more about how effectively they can control them.