AI makes governance a boardroom issue
As AI becomes embedded in business processes and contributes to decisions, responsibility cannot remain solely with technology teams. Boards need to understand why AI is being introduced into a process, what data it relies on, how risks are assessed and where human oversight remains necessary.
Adela Jansen, Independent Non-Executive Board Director, sees this as an extension of the board’s existing governance responsibility. Boards need to ask how technology connects with business judgement, whether controls are sufficient and how AI-related risks should be tested. Adriana Matache brings the same principle into insurance, where inaccurate recommendations or automated errors can create immediate operational and reputational consequences. For both perspectives, human validation remains central when decisions affect customers, employees or financial outcomes.
Governance is therefore more than a compliance exercise. Done well, it creates the conditions for organisations to scale AI with greater confidence, particularly as the EU AI Act brings risk classification, transparency and accountability into the operating model.
AI is also becoming a question of resilience
AI dependency introduces a new strategic dimension. As organisations build processes around global platforms, models and infrastructure, boards also need to consider what happens if access to a critical technology changes or disappears. Radu Hanga, President, Bucharest Stock Exchange, connects AI implementation with geopolitics and asks how an organisation would respond after investing heavily in a technology that later becomes inaccessible. He also calls for greater transparency when AI contributes to materials used in corporate decision-making. From an investment perspective, Andrei Dudoiu warns that a business model built around a single AI platform may be exposed to dependencies it cannot control.
The issue goes beyond software. Sorin Elisei, Board Member, OMV Petrom, highlights the physical infrastructure behind AI, including computing power, energy, water and data centres, and the competitive implications of access to those resources. He also stresses that expected efficiencies need to be weighed against implementation costs, information security and the continuing need for specialist validation.