Corporate boards face the challenge of governing artificial intelligence
A company has decided to use artificial intelligence systems to select job candidates. While this promises to speed up the screening process, it raises a strategic question: how can we prevent the technology’s embedded criteria from replicating biases present in historical data?
Why has AI governance become a boardroom priority?
The expansion of automated resources has brought responsibilities that extend beyond technology departments. Intelligent tools now influence decisions on credit, hiring, customer service, and performance evaluation. A 2025 global survey by Deloitte, conducted with directors and executives from 56 countries, found that 66% of respondents believed their boards had limited or no knowledge or experience in AI. This data highlights a gap in preparedness among professionals responsible for overseeing corporate risks.
AI governance refers to the set of policies, responsibilities, and control mechanisms that guide the development and application of these systems. Concerns about the technology’s risks are already part of senior leadership discussions. According to PwC’s 2026 Annual Corporate Directors Survey, 71% of directors surveyed believe it is necessary to strengthen their boards’ competencies in artificial intelligence. For executives, ethics and governance have become essential skills to justify investments, oversee systems, and take responsibility for their consequences.
What risks do executives need to be aware of?
Key risks include the misuse of personal data, discriminatory decisions, and incorrect responses generated by generative AI models. For example, an algorithm used in recruitment may replicate patterns of inequality present in the data used to train it. Legislation must also be considered. In Brazil, the General Data Protection Law (LGPD) imposes obligations for the processing of personal information, including when carried out with AI support. Compliance with these rules requires understanding what data is used, its purposes, and the procedures adopted to protect it.
How can controls be established for AI use?
The U.S. National Institute of Standards and Technology (NIST) has developed a risk management model that organizes companies’ actions into four functions: govern, map, measure, and manage. The proposal involves defining responsibilities, identifying potential impacts, evaluating system performance, and monitoring risks during use. For leadership, this means establishing who authorizes the use of each tool, what tests should be performed, and how problems will be communicated.
How can knowledge in ethics and governance be developed?
Executive training can include fundamentals of AI, data protection, risk assessment, and criteria for overseeing automated decisions. Understanding governance norms and models helps formulate questions about suppliers, data quality, and internal responsibilities. Learning must be connected to the business, enabling the evaluation of when an application brings benefits and what controls are necessary. Competence becomes a differentiator when executives can present to boards evidence about the safety, results, and risks of systems in use within the organization.








