Artificial Intelligence (AI) tools and processes are becoming increasingly pervasive in many industry sectors and in many phases of business. AI use is also spreading to corporate processes and functions. For example, as I recently noted, some companies many be using AI tools to draft the MD&A in their periodic reports. And, at least according to a recent post in the Harvard Law School Forum on Corporate Governance by lawyers from the Debevoise law firm, some corporate boards may be “adopting AI meeting tools to assist with the drafting of board and committee minutes.” As the memo’s authors note, board adoption of AI tools for these purposes may offer some benefits, but the use of the tools also entails risk, which board members will want to take into account in using the tools.
The June 23, 2025, law firm memo, entitled “AI Can Draft Board Minutes – But Should It? Considerations for Public Companies,” can be found here.
The law firm memo opens by noting that AI meeting tools offer “several practical benefits,” but also observing that “evaluating the potential risks associate with the use of these features is crucial from a risk oversight, governance, and controls perspective.” The memo outlines key considerations that should be “top-of-mind” for public companies that are considering using AI applications in board and committee meetings, many of these key considerations having to do with confidentiality and accuracy.
Boards considering using AI tools will want to be sure to confirm with the AI provider that company data will remain confidential and will not be used to train any AI model; that humans at the AI firm will not have access to company data; that the AI firm will not share the company data with any other parties absent specifically agreed circumstances; and the AI provider has an effective cybersecurity program sufficient to protect company data.
There are some practical considerations as well. For starters, in many U.S. jurisdictions, the consent of all parties is required to lawfully record meetings. Participants should in any even be given advance notice that a transcription or summary of the meeting will be generated and available for review. Board members will want to be aware that AI-generated transcripts and summaries may contain inaccuracies and should not be considered final. The board secretary will want to review and revise the transcripts prior to circulation and entry into the company’s official records.
There are circumstances where the use of the AI tools may not be appropriate – for example, if the meeting will involve privileged or highly confidential information (ongoing litigation, strategic initiatives, or sensitive regulatory matter). Dissemination of AI-generated materials that may contain privileged or highly confidential information should be restricted to the meeting participants and others within the scope of the privilege or confidence.
Board use of AI tools also entails certain process considerations. For example, boards adopting AI tools will want to consider how AI-generated materials will be stored and how they are circulated, as well as an appropriate retention schedule. Companies adopting board AI tools will aso want to consider whether any AI-generated material may contain information that is subject to any litigation holds, and how that information is to be preserved and reviewed for discovery. Finally boards will want to ensure that the company information controls appropriately restrict access to and prevent the impermissible disclosure of sensitive information.
The memo concludes by noting that boards should not only be aware of these kinds of risks but should also “establish policies, procedures, and effective controls to mitigate such risks.”
It should come as no surprise that the use of AI tools should make its way into the board room. There is a certain sense of inevitability about this development. The concerns that the memo’s authors raise are important. It will only be a matter of time before AI-generated minutes and transcripts are the subject of pre-litigation books and records requests, and from their become the bases of allegations in plaintiffs’ complaints in corporate litigation. This is one key context in which the accuracy of the AI-generated materials will become crucial.
I have a particular concern here, and that has to do with the kinds of allegations plaintiffs’ lawyers raise in “duty to monitor” type cases. The plaintiffs lawyers will use books and records requests to obtain board minutes and will scour the records to see the extent to which the minutes show that the board discussed a “mission critical” topic. The risk is that the minutes do not show the topic being discussed, allowing the plaintiffs’ lawyers to make the argument that “the board didn’t even discuss” the critical topic. The possibility of overly terse board minutes omit discussion of key topics is always present. I fear that with AI-generated minutes this risk is increased. Even if humans review the minutes, they may not spot the omission (as it is always harder to spot an omission than an error). This type of litigation risk highlights the need for heightened vigilance with respect to board use of AI tools.
In my view, this litigation perspective should inform all decisions concerning AI board tools. Boards will want to be sure they are in the appropriate position to defend themselves in the event that litigation later arises.
Special thanks to a loyal reader for sending me a link to the law firm’s memo.