AI governance is tested when business pressure collides with concerns about a system, Amaka Ibeji argues, urging boards to check whether staff warnings can reach leaders and change decisions.
Writing in an Oct. 23 opinion column for BusinessDay, Ibeji, founder of digital trust and assurance company PALS Hub, points to executive backing, money already invested, growth expectations and competition as forces that can weigh on decisions. She does not identify a company, AI system or incident, and the column does not measure how often those pressures shape oversight.
Her board-level questions center on whether employees can speak up safely and whether concerns make their way to someone empowered to respond. She also urges directors to consider whether evidence can displace an investment narrative built on claims of transformation, speed or competitive advantage, and whether approval of an AI system can be reconsidered. “Governance culture determines how they are actually made when pressure enters the room,” Ibeji wrote.
Ibeji uses examples of a system falling short, a senior executive already backing it and competitors moving fast to frame the pressure. These are scenarios in her argument, not a report of a named incident.
The U.S. National Institute of Standards and Technology (NIST) issued its AI Risk Management Framework 1.0 in 2023, organizing risk work into four functions: Govern, Map, Measure and Manage. NIST says governance informs the other functions and risk management should continue over an AI system’s full lifecycle. It also says varied perspectives can help identify risks.
The Organisation for Economic Co-operation and Development (OECD) adopted its AI Principles in 2019 and updated them in 2024. The principles include accountability, traceability and ongoing risk management across the AI lifecycle. They also call for safeguards that let people intervene in, repair or safely retire systems that risk undue harm or behave in unwanted ways.
The NIST framework and OECD principles outline processes and safeguards, but do not show whether a particular company follows them. Ibeji’s column offers board questions, not data on how often organizations meet them.
