.png)
Zango AI has today published The Future of AI Governance & Compliance in Financial Services - a research report drawing on in-depth interviews with 27 C-suite and senior leaders across UK and European financial institutions, supported by four focus groups and four industry roundtables involving a further 60 senior practitioners.
The report was developed with academic input from independent researchers at the University of Glasgow and the Oxford Martin School AI Governance Initiative, and with Dean Nash, Global Chief Operating Officer (Legal) at Santander, serving as industry adviser.
The foreword to the report is contributed by Lord Tim Clement-Jones, with expert commentary from Rt Hon John Glen MP, a member of the UK Treasury Committee. Industry contributors include senior leaders from Revolut, Monzo, Santander, Standard Chartered, Lloyds Banking Group, Stripe, Commerzbank, Allica Bank, St. James's Place, and ICA.
The financial services industry is deploying AI faster than it can govern it
Across banking, fintech, payments, wealth management, and digital assets, AI adoption is accelerating rapidly. In the UK alone, at least three quarters of financial firms are already using AI in some capacity. Yet the functions responsible for overseeing that deployment - risk, compliance, and audit - are falling behind, with limited visibility over what is being used and insufficient tools and skills to govern it effectively.
Two structural gaps are emerging
The research identifies two key governance failures.
The first is a capability gap: oversight functions often lack the skills, tools, and infrastructure to govern AI at the pace and scale of business deployment. Many compliance and risk professionals cannot interrogate how AI systems actually behave, which means independent challenge risks becoming independent observation.
The second is an implementation gap: no authoritative, sector-specific standard currently exists in the UK or EU that translates high-level regulatory principles into operational AI governance practice. Firms are solving the same governance problems independently, without shared standards or a common baseline to work from.
Why this technology is different
The report documents how generative AI and emerging agentic systems present qualitatively new governance challenges. Unlike earlier models where the same input produced the same output, generative AI produces context-dependent results with no single correct answer to test against. Agentic systems - which can act autonomously and at scale - are beginning to move from pilot to production. Governance frameworks designed for an earlier era of technology are under growing strain.
Practitioners interviewed for the research were candid about the risks of inaction. Conduct failures that once took years to accumulate could now occur in weeks. Criminal organisations are already deploying AI to exploit governance gaps at scale. And across the research, a consistent theme emerged: practitioners expect regulatory frameworks to be shaped less by proactive standard-setting than by something going wrong first.
Towards shared AI governance implementation guidance
The report makes a clear call to action. The US and Singapore have already moved to fill the implementation gap through public-private collaboration. The UK and EU have not. The report argues that practitioner-built, sector-specific AI governance guidance - developed with regulator engagement and modelled on what the Joint Money Laundering Steering Group has delivered in financial crime - is both achievable and necessary.
Zango AI convened a roundtable in the House of Lords, bringing together contributors to the research to focus specifically on the implementation gap the report identifies. The session, hosted with the support of Baroness Kingsmill - ex-Chair of Monzo- rallied participants around a shared next step: to collectively build a financial services-specific governance layer for AI agents, establishing the common operational standards the sector currently lacks.
Download the full report
The full report is available now.
