FSB finalizes AI sound practices, with stronger human oversight for agents
The Financial Stability Board published its final 12 sound practices for responsible AI adoption by financial institutions on October 15, finalizing the June consultation. The final version strengthens the agentic AI guidance: in higher-risk uses, more intensive human oversight is critical, and firms should set clear, risk-based triggers for when an agent must seek human approval before proceeding.
The FSB also dropped the draft framing of agents as synthetic employees, noting that AI agents do not reason like humans and may interpret predefined goals in unexpected ways. The practices cover organization-wide AI governance and risk management across the AI lifecycle, including human oversight, cyber and ICT risk, and third-party dependencies.