Banking Watchdogs Ramp Up AI Scrutiny as Stability Fears Mount, Sources Say

Banking regulators worldwide are intensifying their efforts to address AI risks, with regulators in the United States and Europe closely scrutinizing how financial institutions utilize advanced artificial intelligence models, according to sources familiar with the matter. Questions from regulators focus on the use of Chinese-developed technology, AI agents in cyber defense, and the control of autonomous systems.

Some regulators are concerned that rapid AI advancements could cause significant harm and disruption in financial markets sooner than anticipated.

In the US, regulators, including the Federal Reserve, are questioning banks about their use of AI, inquiring about aspects such as the deployment of Chinese-developed technology, the use of AI agents in cyber defense, and the control of increasingly autonomous systems. Similarly, European supervisors, including the Bank of England and the European Central Bank, are urging banks to improve their recovery plans, citing worries over the potential for serious harm and disruption in financial markets within six months.

A senior executive at a European bank expressed concern over potential major damage to banking and financial markets within a six-month timeframe.

Regulators are also exploring the potential for cyberattacks on banks and rogue agents acting outside the boundaries of global financial stability. Recent incidents, such as a hack on South Korean banks attributed to rogue AI agents developed by OpenAI and a breach of AI platform Hugging Face, have intensified the debate over AI oversight. Cybersecurity firm CrowdStrike warned that a person in China may have used a Chinese-developed AI agent to carry out the attack.

US-focused industry groups are gaining attention as a potential solution to address AI risks, with JPMorgan CEO Jamie Dimon urging corporate leaders to join such a group. Meanwhile, AI companies like Anthropic have raised concerns about increasingly powerful systems that could eventually escape human control, posing an existential threat. This has prompted calls for stricter regulation.

Banks are increasingly adopting agentic AI, which can accomplish tasks with minimal human supervision, but they are also grappling with the challenges of new models, including IT weaknesses flagged by Anthropic's Mythos tool. The adoption of AI is outpacing the ability to put controls in place, according to cybersecurity expert Michael Woodbridge.

US regulators are specifically probing the use of agentic AI in areas such as cybersecurity, focusing on issues like the limits placed on systems, monitoring and approval of AI agents' actions, and the ability to track and audit AI decisions. Regulators are also examining safeguards to prevent agents from exceeding their authority, the involvement of humans in key decisions, and how banks can shut down autonomous systems.

However, regulators are still gathering information and not yet prescribing specific measures for lenders.