Overview and Key Practical Takeaways
Artificial Intelligence (AI) in the boardroom is a double-edged sword. It presents risks, but also potential rewards. Meeting their fiduciary duties to the company requires that directors get this balance right.
A new Nasdaq study, a recent Australian court ruling, and a Delaware dispute provide helpful context. We discuss:
- Recent statistics on AI in the boardroom.
- Approved vs unapproved AI in the boardroom and the risk of “shadow AI.”
- AI in (1) preparing board materials, (2) recording board meetings, and (3) strategic board decision-making.
Our key takeaway for boards in Canada? Overreliance on AI tools risks breaching the directors’ duties to the company. However, given AI’s undeniable utility, under reliance on AI could plausibly raise similar risk.
To discuss further, contact any of the authors.
This is Episode 1 in our ongoing AI in the Boardroom series. For more Fasken corporate governance thought leadership, visit our Capital Markets and M&A hub and subscribe.
Recent Statistics on AI in the Boardroom: The May 2026 Nasdaq Report
A study published by Nasdaq in May 2026 states that the use of AI in board work is expected to “expand significantly in the near term.” While only 10% of survey respondents said they were currently using AI in connection with board work, 46% said they were exploring, testing or planning on doing so. On the other hand, 40% stated they were not currently planning on incorporating AI into board work.
What explains these mixed messages? Nasdaq concludes that 56% of respondents leaning into AI in the boardroom reflects interest in improving efficiency in governance matters. The 40% of respondents that have been reluctant to bring AI into the boardroom “reflects concern about AI-related content accuracy, legal risk exposure, and confidentiality.”
Approved vs Unapproved AI in the Boardroom and the Risk of “Shadow AI”
The Nasdaq study confirms what senior executives already know well: AI is quickly permeating every aspect of modern life and business, and the boardroom is no different.
But the study also exposes a key practical distinction: not all use of AI in or around the boardroom is necessarily occurring through approved channels. While the formal adoption of AI in the boardroom appears to be happening incrementally, the survey’s respondents indicate that the use of AI by executives outside of sanctioned channels – so called “shadow AI” – is growing much faster. Specifically, the survey states that while only 8% of directors and executives reported that their board uses an approved AI tool, “by some accounts, approximately two-thirds of board members are currently using AI in their roles as directors.”
Any discussion of AI in the boardroom and the associated legal issues raised must therefore account for both approved and shadow AI use. These discussions must stress, as the Nasdaq survey does, that use of unapproved AI tools in corporate governance “presents heightened risks in terms of accuracy, reliability, confidentiality, and litigation risk.”
What Directors’ Duties Apply to AI in the Boardroom?
AI use by executives, whether approved or unapproved, intersects with directors’ legal duties on two main fronts: the duty of confidentiality and the duty of care. The former requires that directors not disclose any sensitive or proprietary information of the company to third parties. The latter requires that directors exercise the care, diligence and skill that a reasonably prudent person would apply in comparable circumstances.
Closely related to the duty of care is the business judgment rule. This shields directors from liability where they have applied an appropriate rigour in reaching their decisions. However, business judgment must be shown for directors to rely on the rule. Directors also benefit from a statutory good faith reliance defence that allows them to rely on “a report of a person whose profession lends credibility” to their work.
Practical Examples of AI Risk and Reward in the Boardroom
AI in the boardroom raises considerations under each of these corporate law principles. To illustrate, we focus on three key practical examples. These are:
- AI in preparing board materials.
- AI in recording board meetings.
- AI in strategic board decision-making.
While we address these examples separately, they are in practice interconnected, particularly as relates to the duty of care, the business judgment rule and the good faith reliance defence. Also, while our discussion focuses on directors, it applies equally to executives and officers.
AI in Preparing Board Materials
The Nasdaq survey states that the three most common uses of AI in board work are:
- Drafting executive summaries – 67%.
- Summarizing documents or background materials – 65%.
- Developing presentations – 45%.
We are unaware of any Canadian court or regulator having yet given meaningful guidance on the appropriate (or inappropriate) use of AI in preparing board materials. However, the March 2026 ruling of the Federal Court of Australia in ASIC v Bekier specifically addressed this point. The court also recognized the doubled-edged sword of AI in the boardroom.
The court found serious breaches of the duty of care by a company’s former CEO and General Counsel in connection with money laundering at one of the company’s casinos. The use of AI was not central to this finding. That said, certain of the company’s directors had based their defence in part on “information overload”, i.e., that, given the extensive materials the board had routinely been given for board meetings, they could not reasonably have been expected to “read and absorb” all the information supplied. This led the court to comment on the “profound impact” AI is having on corporate governance.
Regarding AI adoption, the court stated that “boards are cautiously but increasingly experimenting with AI as a governance support tool.” The court also flagged “shadow” AI use, noting that it “would be jejune to deny that many individual directors are using AI informally to prepare for meetings.” For the court, it was “contemporary reality” that public company directors “are now commonly” using AI to help them navigate board materials.
Regarding the interaction of AI tools and the board’s duties, the court stressed that “ethical reasoning rests with directors, not machines.” As such, while “AI is already changing the way directors receive and analyse materials”, the onus remains on directors to “ensure this occurs in a responsible way.” The court explained:
[D]irectors must be furnished, by whatever means are adopted, with information in a form that is both comprehensive and capable of proper digestion. It scarcely requires emphasis to observe that the use of AI-generated summaries as a substitute for the careful reading and interrogation of board materials would warrant caution, not least because inadequately deployed or misdirected AI may increase risk and legal exposure rather than mitigate it.
The court expressly acknowledged there is “considerable potential for AI, if appropriately utilised, to assist directors in the discharge of their duties.” That said, the court stressed that the “use of technology may assist comprehension, but it cannot displace judgment.”
AI in Recording Board Meetings
According to the Nasdaq survey, only 18% of respondents stated they were using AI for notetaking at board meetings and only 15% stated they were using AI for transcribing or summarizing board meetings.[1] There has nonetheless been much recent industry talk of the pros and cons of using AI to record board meetings.
The attraction of doing so is the potentially more efficient preparation of board minutes. AI tools can identify different speakers, summarize key decision points, track action plans, and conform the minutes’ style with company precedent and preferences. A task that might otherwise take the corporate secretary hours to complete could be finished within minutes of the board meeting, allowing the minutes to be approved contemporaneously.
That said, several drawbacks remain, and these presumably explain the relatively low adoption rate to date identified in the Nasdaq study. These include:
- A potential “chilling effect”: Recording meetings can have a psychological impact on participants, prompting self-censoring or otherwise undermining the full and frank discussion and debate needed for directors to comply with their fiduciary duties.
- Concerns around accuracy persist: Human transcription may better understand tone and nuance and better contextualize discussions, debate and decision-points. AI tools, by contrast, may “hallucinate” content that was never said or misunderstand ambiguous, technical or context-dependent discussion. An inaccurate record of board deliberations could itself undermine, rather than support, the directors’ ability to demonstrate they discharged their duty of care.
- Privacy law: To comply with privacy laws, it may also be necessary to obtain the prior informed consent of meeting participants to the use of AI recordkeeping tools.[2]
- Risk of data breaches: Third-party AI tools introduce cybersecurity risk, with board-level discussion (e.g., strategic direction) being of particular interest to malicious actors.
- Potential loss of privilege: Board level matters can include discussions protected by solicitor-client privilege or litigation privilege. These can be waived by disclosure to third parties via AI tools. Appropriate contractual safeguards must be put in place and maintained.[3]
- Commercially sensitive information: As with the potential loss of privilege, use of AI tools risks the inadvertent disclosure of confidential and/or commercially sensitive information to third parties.
An overarching point for directors to appreciate is that, should it ever be necessary, board minutes will be vital to demonstrating they met their duty of care and were sufficiently diligent to attract the protection of the business judgment rule and/or the good faith reliance defence. And so, even if AI is used to record board meetings, a “human in the loop” will remain critical to ensure a context-appropriate transcript.
AI in Strategic Board Decision-Making
The most common concern cited by respondents (79%) in the Nasdaq study in using AI in board work is the “accuracy and reliability of AI-generated content.” As just discussed, this is presumably a driver of the relatively low adoption rates of AI tools in recording board meetings to date. This issue also presents three interrelated problems in connection with board decision-making, the duty of care, the business judgment rule, and the good faith reliance defence.
First, directors should ensure that AI-assisted information they are relying on is accurate and trustworthy. This requires sufficient AI literacy among board members to be confident AI output has been adequately vetted and stress tested. This in turn requires that the board has well-founded confidence in the AI literacy of those persons preparing AI-assisted materials and analysis for the board’s consideration.
Boards generally cannot reliably build this AI literacy on their own. Given the complexity of AI and how quickly it is evolving, genuine literacy may be best imparted by specialized outside experts. Engaging qualified external advisors may also strengthen the directors' position vis-à-vis their duty of care and under the good faith reliance defence.
Second, directors should ensure they are bringing independent human reasoning and wisdom to bear on important matters. This must remain demonstrable. Strategic decision-making cannot be outsourced to AI. AI tools can be used as a sounding board, to make counterarguments, to propose alternative approaches, and to deepen the board’s debate. But, as cautioned by the court in ASIC v Bekier, the duty of care “remains personal, and it requires informed human judgment.”
Lastly, shadow AI raises particular risk on this front. A critical function of official AI policies is to ensure AI inputs into board work are transparent and controlled, including so such inputs are adequately vetted and stress tested. Shadow AI use threatens to covertly inject AI reasoning into board analysis and/or strategic assessment without the appropriate guardrails having been applied. Should such shadow AI reasoning infiltrate the board’s strategic decision-making in a material way, the duty of care may be breached, and the protection of the business judgment rule may be lost. Other unintended and adverse consequences could also result. By way of example, in an ongoing M&A dispute in Delaware, the sellers have, in making their breach of contract claim, relied heavily on evidence of exchanges between a senior executive of the buyer and an AI chatbot regarding corporate strategy and how the buyer might be able to avoid its performance obligations.
Key Practical Takeaways for Directors, Executives and General Counsel
Our first key takeaway for directors is that AI in the boardroom is a double-edged sword. Overreliance on AI tools risks failing to meet the directors’ fiduciary duties to the company. It also risks failing to secure the protection of the business judgment rule and good faith reliance defence. However, given AI’s undeniable utility, under reliance on AI could plausibly raise similar risk from the opposite direction in respect of the board’s duty of care: boards should not unduly avoid technology that can enhance the depth and vigour of their decision-making.
Second, to get this balance right, it should be appreciated that AI can manifest in the boardroom in multiple different ways. Corporate AI policies and procedures should consider the different applications of AI to board work altogether and not in isolation. Boards should ensure sufficient AI literacy among those persons preparing AI-assisted materials and analysis for the board’s consideration, including to ensure they have been adequately vetted and stress-tested. The board, on the other hand, must ensure they apply independent human reasoning and judgment to the issues before them. To this end, boards should adopt internal AI policies and guidelines tailored to their organization’s particular circumstances and risk profile and ensure these are consistently followed in practice.
Third, the greatest AI-related risk arguably arises from shadow AI. An area of acute concern for boards should be the potential for the inadvertent disclosure of confidential information, competitively sensitive information and, in the context of public companies, material non-public information. Potential adverse effects of AI on well-informed and independent decision-making should likewise not be overlooked. The potential for AI “hallucinations” is well known. But other and less easily detectable risks are also at play. These include the sycophantic nature of certain AI tools, i.e., a tendency for the AI tool to affirm the AI user’s assumptions or preferences or otherwise tell the AI user what the AI expects the AI user may want to hear. This reinforces the need to vet and stress-test AI analysis and the general notion that AI better serves as a sparring partner than a coach or trainer.
The authors thank Léon-Sékou Tétreault, student at law in our Montreal office, for his assistance in the preparation of this article.
