Artificial Intelligence Consortium minutes – June 2026
Summary
The Bank of England’s AI Consortium discussed governance, explainability and model risk management for generative AI in financial services. The minutes highlight concerns around third-party dependencies, unpredictable outputs and the limits of existing model risk frameworks, alongside proposals for more outcome-based validation and stronger oversight.
Why it matters
While focused on financial services, the discussion is relevant to surveyors because AI tools are increasingly used in valuation, risk assessment and workflow automation. The emphasis on governance, transparency and third-party controls is also pertinent to firms adopting AI-enabled property and compliance processes.
Key points
- GenAI creates explainability and transparency challenges for existing model risk frameworks.
- Third-party providers and interconnected systems can reduce visibility and increase operational risk.
- The workshop favoured managing AI at the full system level, not just individual models.
- Members discussed outcome-based validation, human oversight and auditable documentation.
- AI adoption may alter contagion pathways and amplify system-wide disruption in stress periods.
This is an RPSA summary of a publicly available article. The full content remains with the original publisher.
