Artificial intelligence (AI) is rapidly becoming part of how we write, analyse and decide. The question is no longer simply whether AI is used in professional work, but how materially it contributed, and who is accountable for the judgment that follows.Universities have begun wrestling with this. Cambridge, Sydney and Monash all provide for acknowledgement of permitted AI use. At Wits, where I teach, the university-wide AI policy requires transparent disclosure and stresses AI should augment, not substitute for, human intellect.The principle is more useful than a blanket declaration: where AI use is permitted, its material contribution should be visible. That raises an uncomfortable question. If a student may be expected to declare AI assistance in a 2,000-word essay, should an executive not consider similar transparency when preparing a R200m recommendation for a board?AI is already in the roomIn most boardrooms AI is already present, but not always visible. Executives use it to summarise reports, refine strategy documents and test assumptions. Directors can use it to interrogate papers or explore scenarios.The genie is not merely out of the bottle. It has probably already helped prepare the PowerPoint. This should not alarm us. Used intelligently, AI can improve thinking. A CEO who asks AI to expose weaknesses in her strategy may be exercising better judgement, not worse.The governance issue is therefore not whether AI should be used. It is whether AI influence over analysis and judgment should remain invisible.SA has had two warningsThis is no longer abstract. In April the department of communications & digital technologies withdrew its draft national AI policy after fictitious references were confirmed. Minister Solly Malatsi said the most plausible explanation was unverified AI-generated citations and called the episode a lesson in vigilant human oversight.Days later, the department of home affairs suspended two senior officials over apparent AI hallucinations in its revised immigration white paper. Its response was particularly relevant: AI checks and declarations would be introduced into internal approval processes.The lesson is not to stop using AI. It is to know where AI entered a process, what it contributed and who checked the result.From outputs to provenanceInternationally, the same themes are emerging. The Organisation for Economic Co-operation and Development’s AI principles emphasise transparency, disclosure, accountability and traceability; the EU AI Act introduces transparency obligations for specified AI systems and AI-generated content.None of this means every board memorandum touched by a large language model (LLM) needs a warning label. However, increasingly we need to understand not only an output, but something about how it was produced.For boards that concern might usefully be called AI provenance: visibility into origin and chain of custody — where information came from and how it was transformed.In the boardroom that means knowing where and how AI materially contributed while preserving clear human accountability for the final judgment.Assistance is not substitutionNot all AI use is equal. One executive uses AI only for readability. Another uses it to research alternatives and test assumptions. A third asks AI what strategy the company should pursue and substantially incorporates its recommendation.All three could truthfully tick “AI used: Yes”. But their governance implications are not the same. We need to distinguish AI-assisted work from substitution of human judgment. The former may signal professional competence; the latter should make a board much more interested in how the recommendation was formed.Boards govern judgmentBoards don’t merely consume information; they rely on the judgment behind it. When a CEO recommends a major investment or acquisition, directors are testing management’s assumptions and willingness to stand behind the recommendation.If a director challenges a critical assumption and the executive cannot explain it because the analysis was substantially generated elsewhere, we have something peculiar: an impressive answer without an identifiable thinker.Executives have always relied on advisers and strategy teams. AI is not fundamentally different. Governance still requires clarity about who owns the recommendation.AI should not become the first adviser in corporate history that everyone consults but nobody acknowledges.What is good for the student...Here lies the irony. Universities teach students about attribution, hallucinations, verification and disclosure. Those students then enter organisations where more consequential decisions may be made with the same technology, often without equivalent expectations.A student may have to declare using an LLM for an assignment. An executive might use one to help formulate a multimillion-rand investment proposal and say nothing. That asymmetry is difficult to defend.Perhaps professional life needs its own version of academic integrity because both business and universities depend on intellectual accountability.King V gives us the spaceSouth Africa does not need another code prescribing every technological contingency. King V continues our principles-based approach to corporate governance.Accountability, transparency, ethical leadership and effective control must be translated into practices appropriate to changing circumstances. Material AI use may now be one of them.The question is not whether King V specifically requires AI disclosure, but what good governance requires when technology influences the information and judgment on which a board relies.Perhaps just one small boxThe answer may be remarkably mundane. Board paper templates already contain fields for the sponsor, decision and risks. Perhaps they need one more: Material AI use: Yes/No. If yes: Purpose — Research/Analysis/Synthesis/Scenario development / Drafting/Recommendation support/Other.Management has reviewed and validated the work and accepts full accountability for the conclusions and recommendations contained in this paper.Nobody needs to declare that AI corrected two commas. The threshold is materiality: did AI meaningfully contribute to the reasoning or recommendation the board is being asked to rely on?New governance norm?Boards traditionally ask directors to declare conflicts of interest. Perhaps the AI era requires a broader understanding of what deserves disclosure.The question should not become “Did you use ChatGPT?”. The better question is: “Did AI materially influence the analysis or judgement before us, and who is taking responsibility for the result?”In the spirit of that proposition, I should disclose my own provenance. I used a paid LLM as a sounding board on specific parts of this article to test the argument, refine drafting after my own writing, and for reference checking. The views, judgment and final responsibility are mine.We are still writing the rules of this game. If we expect students to disclose material AI assistance, perhaps graduates should do the same, not because AI is something to confess, but because transparency about how judgment is formed is what good governance has always been about.Pillay is a non-executive director, governance practitioner and faculty member at Wits Business School (WBS). He has a particular interest in AI and emergent technology governance and multi-stakeholder partnerships. He writes in his personal capacity.