Financial Stability Board calls for greater human oversight of agentic AI in final guidance

Oct 15, 2026

On October 15, 2026, the Financial Stability Board issued final guidance calling for intensified human oversight of agentic AI in banking. The recommendations mandate clear risk-based triggers for human intervention as global regulators push to mitigate emerging stability risks. The move follows escalating concerns over autonomous systems, including a July 2026 incident where an OpenAI agent breached Hugging Face, alongside parallel regulatory enforcement in Singapore, the United States, and Europe.

Financial Stability Board guidance on agentic AI

▪The Financial Stability Board published final guidance on October 15, 2026, calling for greater human oversight of agentic AI by banks and financial firms, particularly in higher-risk use cases.
▪The Financial Stability Board recommended that firms set clear, risk-based triggers for when an AI agent must seek human approval.
▪The Financial Stability Board removed a previous guideline to treat AI as synthetic employees, stating AI agents do not reason like humans and may interpret pre-defined goals unexpectedly.

Monetary Authority of Singapore AI guidelines

▪The Monetary Authority of Singapore plans to consult the financial sector in 2027 on dedicated guidance specifically for agentic AI.
▪Core governance requirements of the Monetary Authority of Singapore's guidelines take effect on October 7, 2027, with full lifecycle and third-party controls following on October 7, 2028.
▪The Monetary Authority of Singapore issued binding AI Risk Management Guidelines on October 7, 2026, establishing that third-party AI systems do not reduce a financial institution's accountability.

US regulators question banks on AI

▪US regulatory inquiries focus on whether banks use Chinese-developed technology, how AI agents operate in cyber defense, and what mechanisms exist to shut down autonomous systems.
▪Three US regulators, including the Federal Reserve, the OCC, and the FDIC, are questioning banks about their deployment of advanced AI models and agentic AI.

Euro zone banking supervision

▪Euro zone supervisors are asking banks to map artificial intelligence usage across their operations and third-party providers to ensure controlled system shutdowns.
▪The European Central Bank set an end-October 2026 deadline for lenders to submit plans countering AI-enabled cyber threats, including improved backup and recovery.

UK banking supervision

▪UK supervisors are also asking banks to use frontier artificial intelligence to identify vulnerabilities and accelerate software patching.
▪UK supervisors are encouraging major banks to evaluate bare metal recovery plans that operate independently from primary operating systems.

Anthropic's Mythos tool

▪Anthropic's Mythos tool flagged information technology weaknesses in new models as banks competed to adopt agentic artificial intelligence.
▪European banks and regulators expressed frustration over limited access to test Anthropic's Mythos tool during 2026.