AI governance fuels boardroom tension

A recent study by Veeam Software reveals growing tension within boardrooms as organizations struggle to control artificial intelligence (AI) autonomy within their operations, particularly concerning sensitive data. The research, conducted among 1,000 enterprise IT, data, and security decision-makers in the UK, Germany, France, and the Middle East and Africa, reveals that 70% of surveyed organizations have automated AI workflows interacting with sensitive corporate data without proper oversight.

This has led to executive accountability becoming a boardroom risk, with 32% of organizations already experiencing tension or conflict among executives due to rising regulatory pressure and corporate accountability.

The findings highlight a widening gap between the rapid deployment of autonomous AI and organizations' ability to understand and control these systems' interactions with corporate information. Over a third of surveyed organizations (32%) report that regulatory pressure and corporate accountability are creating conflict among executives. This uncertainty is reflected in executive concerns, with 58% of enterprises now under new corporate accountability laws, and 12% uncertain about individual responsibilities.

C-level leaders are also taking personal risks, with 40% of respondents expressing worry about personal liability or consequences of deploying autonomous AI systems. Additionally, 39% noted greater board-level scrutiny, while 37% reported increased personal stress or anxiety due to the heightened accountability. Tim Pfaelzer, GM and senior VP for EMEA at Veeam, emphasizes that executives now face more factors to consider regarding data governance and cyber resilience regulations.

While the tension and stress are understandable, Pfaelzer believes that better alignment at the board level would ultimately improve the organization's ability to trust its data and AI governance in the agentic era.

Organizations are responding to these challenges by altering where and how AI models are deployed. Notably, 41% of EMEA organizations are building their own local or sovereign AI models to address shadow AI concerns, while 49% are adopting hybrid approaches, combining local or sovereign models for sensitive workloads with global models for more general applications.

As the industry grapples with AI governance, the study suggests that the focus should shift from controlling thousands of autonomous agents individually to securing, governing, and understanding the data upon which these agents depend.