The Financial Stability Board (FSB), a global body that sets standards for the financial system, has asked banks and other financial firms to keep more humans in charge when they use “agentic” AI. It published its final guidance on Thursday, 15 October 2026. Here is what it says, in simple terms.
What Is Agentic AI?
Agentic AI means AI systems, often called AI agents, that can plan and carry out tasks with little help from people. A normal chatbot answers a question. An AI agent can go further and take steps on its own, such as moving data, sending messages or making decisions inside a company’s systems.
What the FSB Is Asking For
- More human oversight for higher-risk uses. The riskier the task, the more closely people should watch what the AI agent does.
- Clear, risk-based triggers. Firms should decide in advance when an AI agent must stop and get approval from a human.
- No more “synthetic employee” thinking. The FSB dropped a suggestion in its earlier draft to treat AI agents like synthetic employees.
Why the FSB Dropped the “Synthetic Employee” Idea
The FSB now says firms should remember that “AI agents do not reason like humans and may interpret pre-defined goals in unexpected ways.” In simple words, an AI agent may follow the instructions it was given but reach the goal in a way no person expected. So firms should not assume it will behave like a staff member.
Why the Guidance Changed
The FSB published a draft in June. It said AI agents create distinct challenges for human oversight. The final version was changed after feedback on the draft, growing interest in this technology across the financial sector, and the special risks and controls it involves.
The Bigger Picture
The guidance comes as regulators around the world work harder to limit AI risks. Reports say their concerns include AI-enabled cyber attacks and the chance that AI systems could get around safeguards.
What This Means for Banks and Customers
- For banks: They will need clear rules on which tasks an AI agent can do alone and which need a human to approve.
- For customers: The aim is to lower the chance of costly errors, unfair decisions or security problems when AI handles money-related work.
- For regulators: The FSB sets standards, but it is up to national regulators to turn them into rules in each country.
The FSB guidance is not a law. It tells firms and national regulators what good practice looks like.
Frequently Asked Questions
What is the Financial Stability Board?
It is an international body that sets standards for the global financial system.
What did the FSB say about agentic AI?
It said financial firms should use more human oversight for higher-risk AI agent use and set clear triggers for when an agent must get human approval.
What is a “synthetic employee”?
It was an idea in the FSB’s June draft that treated AI agents like employees. The FSB removed it from the final guidance.
Is the guidance legally binding?
No. It is guidance for firms and national regulators, who decide how to apply it.
Why does human oversight matter?
AI agents may not think like people and may reach goals in unexpected ways, so a human check can catch mistakes in time.
As banks adopt more AI agents, the central message from the FSB is simple: the more risk a task carries, the more a human should stay in control.
