The Financial Stability Board (FSB) is the international body that monitors and makes recommendations about the global financial system. It is hosted at the Bank for International Settlements (BIS) in Basel and coordinates national financial-stability authorities — central banks, finance ministries, and securities regulators — across G20 economies.
Mandate
The FSB's mandate covers cross-border systemic risk, including bank-run dynamics, derivatives infrastructure, payments resilience, and, increasingly, cyber and operational resilience. Its work on the latter includes a "Cyber Lexicon" and operational-resilience guidance.
AI involvement
On May 17-18, 2026, the Financial Times reported that Anthropic agreed to brief the FSB on the capabilities of its Claude Mythos Preview model, specifically its ability to find and exploit software vulnerabilities at scale. The briefing extended Anthropic's earlier private cooperation with European banking regulators, who had asked Anthropic about Mythos-related risk to bank cybersecurity (Source: ft.com). According to the report, it was the first publicly reported case of an AI lab briefing the international financial-stability community on a frontier model's systemic-risk profile.
The engagement extended the FSB's cyber and operational-resilience work into model-specific capability assessment, and ran in parallel with an IMF warning about Mythos in May 2026 (see International Monetary Fund (IMF)). It is one of several reported cases of an AI lab engaging directly with sectoral authorities, alongside the DPA designation in defense and the Vatican engagement in ethics. The FSB engagement situates the FSB alongside other international AI-governance bodies such as the EU AI Office, the UK AISI, and various national AISIs, and connects the systemic-risk discussion of Autonomous cyber-agents to financial regulation.
The FSB moved from capability briefings to supervisory guidance in June 2026. On June 10, 2026 it urged financial firms to treat autonomous AI agents as "synthetic employees," set clear boundaries on agent actions, and require human approval for high-risk actions such as transactions above set thresholds, in non-binding guidance open for feedback until July 22; it cited a Cambridge survey finding 52% of financial-sector respondents already adopting agentic AI (Source: reuters.com). See Agentic AI, Financial Services — AI Deployment. On July 13, 2026, Federal Reserve vice chair for supervision Michelle Bowman urged stakeholders to comment on the FSB's "sound practices" report on responsible AI adoption by financial institutions (Source: insideaipolicy.com).
Relationships
- regulated-by: (international body; not regulated)
- related: Anthropic, Claude Mythos Preview, Autonomous cyber-agents, AI and Cybersecurity, International Monetary Fund (IMF), Financial Services — AI Deployment (industry)