By Even Hashikutuva and Chisom Obiudo
In 2021, Zillow shut down its AI-powered home-buying operation after losing over $800 million. The algorithm kept purchasing homes in the wrong markets at the wrong prices.
Knight Capital’s trading algorithm lost $440 million in just 45 minutes in 2012. Apple’s credit card algorithm, developed with Goldman Sachs, gave women lower credit limits than men with identical financial profiles, triggering a regulatory investigation by the New York Department of Financial Services.
These weren’t IT errors or technology glitches. They were governance failures that reached the boardroom.
Artificial intelligence touches nearly every industry, from banking and insurance to retail, manufacturing, and government services.
Yet most Namibian boards delegate AI decisions entirely to technology teams, assuming it’s a technical problem rather than a strategic one. It isn’t.
Consider what is happening in Namibia’s HR departments today. Organisations are quietly adopting automated screening tools to rank CVs, filter candidates, and recommend hires. Globally, these systems have been shown to amplify bias and exclude qualified talent.
Amazon scrapped its AI recruiting tool after discovering it systematically discriminated against women, penalising resumes that included the word “women’s” or graduates from women’s colleges.
But in Namibia, most boards would struggle to answer a basic question: if an algorithm rejects a candidate, who is accountable? The HR manager? The software vendor? The board itself?
The shift is happening globally, and it’s accelerating. According to a 2025 report by the Ernst & Young (EY) Center for Board Matters, 48% of Fortune 100 companies now formally cite AI risk as part of their board oversight responsibilities, up from 16% the previous year.
Companies are treating AI governance as a boardroom priority because regulators are watching, investors are asking harder questions, and the financial and reputational consequences of AI failures have become too severe to ignore.
In Namibia, this discussion has only recently begun. A Data Protection Bill is expected to be tabled in Parliament this year, followed by a proposed AI Bill aligned with UNESCO’s ethical guidelines and regional frameworks such as the SADC Model Laws.
The EU AI Act already imposes fines of up to €35 million or 7% of a company’s total worldwide annual turnover for illegal AI practices, whichever is higher. Waiting for regulators to act first causes companies to rush and puts them at a competitive disadvantage.
AI risk is already a board duty
Fiduciary responsibility: Under Section 430 of the Namibian Companies Act 28 of 2004, a court can declare directors personally liable, without any limitation of liability, for the debts of a company if its business was carried on “recklessly, or with the intent to defraud creditors.”
In the future, this could be interpreted to include the deployment of high-risk AI without proper oversight.
This risk becomes real when a bank deploys an AI credit scoring model trained on historical data. If that model systematically denies loans to applicants from certain areas while approving similar applications elsewhere, the board is responsible, not the data science team.
In such a case, the court’s inquiry into reckless conduct would focus on the boardroom: Did your board know that this system had been implemented? Did you review its training data? Did you test for bias?
Consider a local scenario: a telecommunications company uses an AI-driven system to determine which customers qualify for device financing.
If the system, trained on data skewed toward urban subscribers, systematically declines applicants in rural areas, many of whom are from indigenous communities, the company could face claims of indirect discrimination under any future data protection or consumer protection legislation.
Regardless of future legislation, the reputational risk alone carries direct commercial consequences.
Risk oversight: Your board already reviews a monthly risk register. Add AI-specific line items immediately: model drift (your AI was trained on 2022 data but it’s now 2026, has performance degraded?), third-party vendor risk (your HR screening vendor in South Africa was just fined for discrimination, are you exposed?), and regulatory breach (can you explain in plain language why your AI denied a loan to a specific customer?).
Strategy and capital allocation: Before approving AI investments, always ask: What’s the expected return on investment (ROI) with measurable outcomes? What’s the deployment timeline? What happens if it fails? Who at the executive level is accountable? These are the same questions you ask about a factory expansion. AI spending requires identical scrutiny.
Five actions for the board
- Establish an AI Oversight Committee. Assign oversight of AI to your existing Risk & Audit Committee or establish a dedicated Technology & AI Governance Committee. The committee’s charter should include: quarterly AI risk assessments; the authority to approve or reject high-risk AI deployments; direct access to the Chief Technology Officer (CTO) and the Chief Risk Officer (CRO); a mandate to commission independent AI audits; and a requirement to report AI incidents to the full board within 48 hours.
- Request a Complete AI Inventory. Ask your CTO, Chief Financial Officer (CFO), and Chief Human Resources Officer (CHRO) for a detailed list of all AI systems in use, in development, or under review. For each system, include the system name and vendor, business function, deployment status, data sources, decision authority (whether autonomous or human-assisted), risk classification, and the accountable executive.
- Set Board Approval Thresholds. High-risk AI (credit decisions, hiring, customer-facing systems without human review, safety-critical applications) requires full board approval before deployment. Medium-risk AI requires executive approval with board notification. Low-risk AI (email filtering, calendar scheduling) requires only CTO approval.
- Ask Five Questions About Every High-Risk System. What data does it use, and what data does it not use? What is its error rate by demographic group? When was it last updated? Who can override it? If you are sued, can you explain this system to a judge?
- Hire an external AI audit service. Engage a reputable auditing firm or qualified local legal counsel to review your highest-risk AI systems. The audit scope should include bias testing across gender, race, language, and geography; model explainability; data privacy compliance; vendor contract review; and incident response preparedness.
There is a stark difference between a company that spends the next six months building an AI governance framework and one that scrambles to assemble one after a regulatory notice arrives. The first sets the terms. The second inherits them
The board meeting is in two weeks. Add AI governance to the agenda.
*Mr. Hashikutuva is a Namibian AI-driven automation strategist, entrepreneur, and co-founder of the tech startup Refrane. Ms. Obiudo is a legal practitioner with specialisation in corporate governance and AI governance.She serves on the National Artificial Intelligence Technical Advisory Committee on law and governance.
