UK FCA Proposes Mandatory Reporting for Significant AI Failures

• •UK FCA calls for mandatory reporting of significant AI failures and near-misses.
• •The initiative aims to enhance regulatory oversight of AI deployments in financial services.
• •Concerns arise about the potential ineffectiveness of compliance penalties.

Jessica Rusu, Chief Data Officer of the UK Financial Conduct Authority (FCA), has called for mandatory reporting of 'significant AI failures and near-misses' during a speech at the Bloomberg UK Policy Series. This initiative aims to enhance regulatory control over AI deployments amid rising concerns about agentic AI misbehavior. Rusu emphasized the need for firms to bolster cyber and operational resilience, stating that current regulations requiring major incident reporting should also apply to AI-related failures. The FCA plans to collaborate with industry and regulators to develop a framework for agentic finance, focusing on accountability and assurance. Pieter Arntz from Malwarebytes defined significant AI failures as those causing serious consequences, while near-misses are events that could have led to harm but were averted. The FCA's approach may face challenges if penalties for non-compliance are not substantial enough to deter negligence.