UK businesses face widening AI governance gap, IBM survey finds

AI-related incidents reached the board or C-suite at 80% of UK organisations surveyed by IBM during the past year, but only 35% changed governance policies or frameworks in response. Incidents included agent failures, data exposure, decision errors and security breaches. Censuswide conducted the research for IBM, including 504 UK respondents at organisations with 1,000 or more employees.

Deployment often continues despite gaps in oversight: 61% proceed within agreed risk limits while governance adapts, while 27% delay deployment until controls are ready. The challenge could intensify as agents gain independence. Although 68% expect agents to have some decision-making autonomy by 2027, only 53% believe their governance structures can scale accordingly.

Formal board responsibility has not resolved accountability. While 97% have assigned AI risk at board level, respondents lack consensus on who should answer for adverse agent outcomes or bear legal liability. Executives identified human escalation thresholds, real-time monitoring, audit trails and contractual vendor liability frameworks as necessary safeguards for increasingly autonomous systems.

Financial and supply chain oversight also remain uneven. Only 50% have a formal, board-reviewed assessment quantifying financial exposure from governance gaps. Meanwhile, 46% acknowledge limitations in coverage of AI used by vendors, technology partners and supply chain participants, and only 19% continuously monitor third-party AI risk.