UK businesses are more excited about AI than ever before — but many are split on where the responsibility lies

A recent IBM study has revealed that UK businesses are more enthusiastic about integrating artificial intelligence (AI) into their operations than ever before. However, the research also highlights a significant divide among business leaders regarding who should be held accountable for AI-related incidents.

According to the survey of over 3,000 business leaders across the EMEA region, 80% of respondents reported that an AI-related incident had reached their board or C-suite in the past year. Examples of such incidents include agent failures, data exposure, AI-driven decision errors, or AI security breaches. Despite this growing concern, only 40% of these organizations plan to modify their AI governance policies or frameworks in response, indicating a potential gap in governance practices for businesses of all kinds.

The survey also found that three-quarters (75%) of respondents expect AI agents to have some level of autonomy with decision-making by 2027. However, only 37% of these executives believe their current governance structures are designed to accommodate this level of autonomy.

While 98% of the surveyed executives have formally assigned responsibility for AI risk at the board level, the majority (26%) have established dedicated board-level AI committees. This suggests that while awareness of the issue exists, there is still much work to be done in terms of establishing liability frameworks for AI agents.

The debate around AI governance is shifting from simple adoption to the crucial matter of accountability. Ana Paula Assis, IBM Senior Vice President and Chair, EMEA and APAC, emphasized that as organizations progress from using AI as assistants to deploying autonomous agents, governance can no longer be treated as a mere compliance exercise. Instead, it has become a business requirement and a value enabler.

To realize the greatest business value, enterprises must scale AI with accountability and trust. This requires building governance into how agents operate, setting clear boundaries on their capabilities, defining clear ownership when things go wrong, and implementing real-time monitoring and audit trails of agent decisions and actions.