• Major US banks are blocking AI agents from accessing customer accounts
• Socure expects clearer industry protocols within approximately six months
• JPMorgan Chase, Bank of America and Citi have restricted agent access
• Executives warn of fraud, cybersecurity, data protection and liability risks
• PNC believes autonomous financial transactions remain years away
• Fifth Third highlights the need for know-your-agent authentication standards
• Socure has launched tools to detect AI agents and verify customer authorization
• Banks face growing pressure to balance innovation with financial security

Major US banks are scrambling to prevent artificial intelligence agents from accessing customer accounts and executing unauthorized transactions, as financial institutions confront growing cybersecurity threats and uncertainty over liability.

Identity and risk intelligence provider Socure believes banks could establish clearer protocols governing AI agents within six months, although some banking executives expect fully autonomous financial transactions to remain years away.

Socure CEO Johnny Ayers said the rapid proliferation of consumer-facing AI agents had triggered urgent discussions across the banking industry.

"There's a lot of code reds going on in the banks right now," Ayers said, highlighting concerns over security vulnerabilities, customer authorization and financial crime prevention.

Socure works with approximately 3,000 banks, including 19 of the 20 largest US institutions, and around 700 fintech companies.

JPMorgan Chase, Bank of America and Citi have blocked AI agents from accessing customer accounts. Nevertheless, reports indicate that some agents have entered banking systems and performed activities beyond their authorized instructions.

Ayers said banks were adopting defensive measures while developing standards governing what automated systems can legitimately do on customers' behalf.

Existing anti-money laundering, know-your-customer and fraud prevention frameworks were not designed to accommodate autonomous agents capable of opening accounts, accessing financial information or initiating transactions.

"Can an agent open a bank account on your behalf? What is it allowed to do? What are its limits?" Ayers asked.

Data protection represents another significant concern, particularly where consumers unknowingly authorize agents to access sensitive transaction information.

Ayers warned that browser-based agents could assume control of customer sessions, making it difficult for institutions to distinguish legitimate users from automated activity.

PNC Financial Services CEO Bill Demchak expressed particular skepticism about allowing agents to execute financial transactions independently.

"We won't let an agent come in and click buttons - in effect, move money on your behalf - without you in the loop, authentication on us, agreeing to the transaction," he said.

Demchak distinguished between AI systems providing financial recommendations and those independently executing transactions, arguing that the industry's liability arrangements remain inadequate.

He suggested banks were "years away" from permitting autonomous agents to control transactional outcomes.

Fifth Third Bancorp has adopted similar restrictions, with executives emphasizing the need for authentication standards capable of verifying both an agent's identity and its authorization to act.

The bank's leadership has also highlighted the potential emergence of "know your agent" requirements comparable to existing customer identification obligations.

Meanwhile, fintech companies are taking more permissive approaches, creating competitive pressure on established institutions to improve their digital services.

Socure has introduced technology designed to detect AI agents, assess their trustworthiness and connect them to verified customers.

The company's new capabilities aim to distinguish legitimate consumer-authorized agents from malicious bots or unauthorized automated activity.

Ayers said banks were deploying detection systems across customer interactions, including account registration, login procedures and call centers.

"It's really every single page, because now I could be logged into my account and then I could have an agent take over my browser," he warned.

However, identifying automated activity creates further questions about whether banks should suspend sessions, block access or permit transactions under specified conditions.

The challenge illustrates a widening gap between rapidly advancing AI capabilities and the authentication, governance and liability frameworks required to deploy them safely across financial services.