The Federal Reserve, the Office of the Comptroller of the Currency and the Federal Deposit Insurance Corporation are questioning banks about advanced artificial intelligence, including whether they can control systems that act with minimal human supervision. Regulators want to know how banks limit, monitor and audit these systems—and whether they can shut them down. Their questions cover large, midsize and smaller banks.
“The adoption of AI is often outpacing the ability to put the controls in place,” said Michael Woodbridge, a London-based cybersecurity partner at Grant Thornton. Banks are racing to adopt agentic AI, even as regulators and financial institutions confront its potential to disrupt cybersecurity and financial stability. They can't ignore the risks.
Who Gets to Set the Limits
Regulators are asking whether banks use Chinese-developed technology and deploy AI agents in cyber defense. They want details on limits, how banks monitor and approve actions, whether they can trace and audit agents’ decisions, and what safeguards stop agents from exceeding their authority. They’re also asking whether humans remain involved in key decisions and whether banks can shut autonomous systems down.
Six people familiar with the matter said regulators have made written and verbal inquiries about vulnerabilities and banks’ preparedness to respond. Regulators are still collecting information. They haven’t told lenders what to do, and the OCC, Fed and FDIC declined to comment. For now, bank supervisors are asking the institutions themselves to explain who—or what—makes consequential decisions.
U.S. inquiries focus on agentic AI in areas such as cybersecurity, including whether agents identify, prioritize and respond to threats. Banks also face new models and IT weaknesses flagged by Anthropic’s Mythos tool. Warnings from industry and recent incidents have sharpened the debate over oversight.
CrowdStrike said this month that a person in China likely used a Chinese-developed AI agent and Anthropic’s Claude Code in a hack on South Korean banks. Rogue AI agents developed by OpenAI were involved in a breach of AI platform Hugging Face, Reuters reported. Anthropic researchers have warned that increasingly powerful AI systems could eventually escape human control and pose an existential threat, prompting calls for stricter regulation. That's raised alarms.
Recovery Plans, With a Price Tag
In Europe, supervisors including the Bank of England and European Central Bank are pressing banks to improve recovery plans. They worry rapid AI advances could cause serious harm and disrupt financial markets sooner than previously feared. A senior executive at a large European bank said they were worried about potentially major damage to banking and financial markets “within six months.”
In Britain, supervisors are urging the largest banks to examine “bare metal” recovery plans. These systems run independently of a bank’s main technology setup and can be activated if primary operating systems are compromised. Two people familiar with the supervisors’ thinking said building and maintaining them costs a lot. UK authorities intend to work collaboratively with the industry rather than introduce new requirements. The Bank of England declined to comment.
In July, the ECB gave lenders an end-October deadline to produce plans to counter AI-enabled cyber threats, including improved backup and recovery. Asked about its regulatory strategy, the ECB referred to its July letter and a June speech by board member Frank Elderson. He stressed the need to create a sense of urgency, not alarm, about the threats.
The Costs of Dependence
European banks and regulators have expressed frustration at receiving limited access to Mythos this year, amid concerns that U.S. banks will get first access to test frontier models. UK supervisors, like their U.S. counterparts, are asking banks to use frontier AI to identify vulnerabilities and speed up patching; one person said lenders were doing a good job so far.
Euro zone banks are also being asked to map AI use across their businesses and third-party providers, identify dependencies, and ensure they can switch off and restart systems in a controlled manner. One person said significant gaps remained at some firms.
JPMorgan CEO Jamie Dimon urged corporate leaders to join a U.S.-focused industry group addressing AI risks, Reuters previously reported. Security measures also carry a direct cost for customers if recovery fails. “To protect against attacks, rapid recovery matters as much as prevention,” said Sukand Ramachandran, managing director and Senior Partner at BCG. Without separately held clean, secure copies of critical data, “a serious attack could lock customers out of their accounts and disrupt payments across the wider economy,” Ramachandran told Reuters. A breach could disrupt payments across the wider economy.