Global financial regulators are grappling with a fundamental problem: they keep issuing overlapping AI guidance that contradicts itself, leaving banks confused about what they actually need to do. The Financial Stability Board (FSB), an international body that coordinates financial policy across major economies, just released a consultation on AI practices for financial institutions, but experts warn it's riddled with ambiguities that could undermine the very risk controls it's meant to strengthen.

Why Are Regulators Creating Conflicting AI Standards?

The FSB published its "Sound Practices for Responsible Adoption of Artificial Intelligence" on June 10, 2026, proposing 12 practices grouped into two categories: organization-wide AI governance and AI life cycle management. The problem is that these practices aren't entirely new. The FSB has been issuing AI-related guidance since 2013, and its core AI risk framework dates back nearly a decade. Over that time, the agency has published multiple reports addressing similar risks but using different terminology and definitions.

The 2017 AI report identified risks like third-party dependencies, market correlations, interpretability gaps, and bias. The 2024 AI Financial Stability Report consolidated these into four categories. Now the 2026 consultation repackages them again as "sound practices," but with new language that doesn't always align with prior guidance. For compliance teams trying to build controls, this creates a moving target.

What Specific Ambiguities Are Causing Problems?

Legal experts analyzing the consultation identified several sound practices that introduce unhelpful confusion rather than clarity. Three practices stand out as having "elevated" ambiguity risk:

• Materiality and Risk Assessment: The consultation references multiple prior FSB documents and Basel Committee guidance, each with slightly different definitions of how to assess which AI risks matter most to a financial institution.
• Explainability and Transparency: The FSB's definition diverges from both the OECD and EU AI Act standards, creating conflicting expectations for how transparent AI systems need to be.
• Performance Management: The guidance restates concepts from national model risk management frameworks but uses different language, making it unclear whether existing controls satisfy the new standard.

The core issue is that the FSB is not a regulator with enforcement power. It's a coordinator that issues non-binding recommendations, which national regulators then adopt, adapt, or ignore. When the FSB's own guidance is internally inconsistent, it creates a cascade of confusion. Banks don't know which definition to follow, supervisors don't know which standard to enforce, and the entire system becomes less effective at catching real risks.

How Are Financial Institutions Supposed to Respond?

The FSB opened a comment period on the consultation that closes July 22, 2026, giving financial institutions a narrow window to shape the final guidance. Experts are urging banks and compliance teams to submit detailed feedback identifying where the sound practices conflict with existing frameworks or introduce unnecessary ambiguity. This is not a bureaucratic formality; the FSB's recommendations typically become influential benchmarks for supervisory examinations once finalized, even though they're technically non-binding.

The stakes are high because AI adoption in financial services is accelerating. Banks are using AI for credit decisions, fraud detection, trading, and customer service. If the regulatory framework is unclear, institutions may either over-comply at great cost or under-comply and miss real risks. Neither outcome serves financial stability.

Steps to Address AI Governance Gaps in Financial Institutions

• Conduct a Mapping Exercise: Compare your current AI risk controls against all relevant FSB guidance (2017, 2024, and 2026 versions) to identify gaps and overlaps in your existing framework.
• Clarify Definitions Internally: Document which definition of key terms like "explainability" and "materiality" your institution will use, and ensure consistency across compliance, risk, and technology teams.
• Submit Feedback to the FSB: If you're a financial institution, use the comment period to flag specific ambiguities and propose clearer language that aligns with your existing controls and industry practice.
• Monitor Regulatory Evolution: Track how national regulators respond to the FSB's final guidance, as they may clarify or reinterpret the sound practices in ways that affect your compliance obligations.

What Does This Mean for the Broader AI Regulation Landscape?

The FSB's struggle with consistency reflects a wider challenge in AI governance: regulators are trying to keep pace with rapidly evolving technology while coordinating across jurisdictions and sectors. The EU AI Act, UK online safety rules, and emerging frameworks in the UAE and other regions are all developing in parallel, often with conflicting definitions and requirements. Financial institutions operating globally face the burden of complying with multiple, sometimes contradictory standards.

Beyond banking, other high-stakes sectors are facing similar governance challenges. Family offices, which manage concentrated wealth and sensitive personal data, are increasingly adopting AI tools but often lack clear governance frameworks to manage the risks. The risks include model drift, hallucinations that create automation bias, data exposure through third-party tools, and AI-enabled fraud. Without clear regulatory guidance, these institutions are left to develop controls on their own, creating inconsistency and gaps.

The FSB's consultation is a critical moment. If the final sound practices resolve the ambiguities and align with existing frameworks, they could provide much-needed clarity. If they don't, financial institutions will continue operating in a fog of conflicting guidance, and regulators will struggle to assess whether institutions are truly managing AI risks effectively. The comment period closes July 22, 2026, and the window for shaping the outcome is closing fast.