2026-08-05 · Source: Bank of England / PRA
Summary in 3 Points • The AI Consortium held its fourth quarterly meeting at the FCA's offices • Workshops focused on explainability, transparency, and model risk challenges in generative AI • Members discussed AI-accelerated contagion risks and the need for cross-firm visibility --- The AI Consortium (AIC) convened its fourth quarterly meeting at the FCA's offices, where discussions centred on the rapid development of AI in financial services. Key topics included the challenges of explainability and transparency in generative AI systems, and the difficulties in applying existing model risk frameworks to these complex systems. Members points to the importance of governance, transparency, and testing of AI model systems, and suggested mitigating factors for more nuanced application of frameworks. The meeting also addressed AI-accelerated contagion risks, proposing a firm-level response taxonomy and stresses the need for cross-firm visibility and engagement with regulators. The potential impact of AI on contagion pathways in the financial system was explored, with members advocating for simulation-based exercises to understand potential failure modes.
For the London insurance market, the discussions on AI model risk and contagion pathways could influence underwriting practices, particularly in assessing the systemic risks associated with AI adoption. Insurers may need to consider the implications of AI-accelerated contagion on financial stability and the potential for correlated errors across firms. The focus on governance and transparency may lead to more stringent policy wording and risk management requirements. Additionally, the focus on cross-firm visibility and regulatory engagement could affect how insurers collaborate with regulators and other financial institutions to manage AI-related risks.