AI Drafting Board Minutes? Hold Up, Wait a Minute. It’s Complicated.

Key Points

Using artificial intelligence tools to record or transcribe board meetings, or generate minutes, can create detailed records of confidential discussions that could create problems if they have to be turned over in litigation.
In addition to the issues with minutes, other uses of AI by directors can pose legal problems. For instance, communications between directors and AI tools are generally not protected by attorney-client privilege, and AI-assisted materials may not qualify for work product protection — meaning candid exchanges with a chatbot, as well as AI-generated drafts, may be discoverable in litigation.
AI-generated materials should be treated like any other records and should be preserved when litigation becomes likely.

__________

Artificial intelligence (AI) tools for board-related work have matured quickly. AI can comb through prior minutes on similar topics, summarize lengthy materials ahead of meetings, and even transcribe meetings in real time and generate draft minutes. All of these things can save time for directors and allow them to focus on certain issues at board meetings.

But the convenience comes with potential pitfalls, and directors need to be alert to them. The stakes are particularly high for AI-generated board minutes, because minutes can turn candid boardroom discussions into a lasting governance record. Many of the same issues also apply more broadly to other uses of AI in the boardroom and in corporate governance.

Using AI to Record and Draft Minutes Is Risky and Could Deter Open Discussion

With all the attention paid to AI at present, directors sometimes ask if AI could be used to compile board minutes or other records. They could, but it would be risky.

Board meetings often cover business updates, strategic planning and legal matters. Directors may speak frankly during those discussions, and disagree, and many of their comments are intentionally omitted from formal minutes. AI-generated transcripts and minutes can capture far more detail than traditional minutes, including unnecessary information, and do not accurately reflect inflections that signal sarcasm or jokes.

Such recordings can potentially chill open director discussion. In addition, AI-produced summaries can contain errors or misleading characterizations.

Another risk to bear in mind is that recordings, transcripts and AI-produced draft minutes would likely need to be turned over to adversaries in litigation, potentially exposing candid, confidential exchanges among directors. This risk is illustrated by the cases discussed below.

AI Communications Are Not Privileged and Are Discoverable

Courts are already being asked to address the issue of discoverability.

Attorney-client privilege generally protects confidential communications between a client and a lawyer made to seek or provide legal advice. But chatbots are not lawyers. A communication between a director and an AI tool does not become privileged just because the topic is legal, the information is sensitive or the vendor agrees to keep it confidential.

Recent court decisions highlight the discoverability risk of AI interactions. A federal court in New York held that an executive’s exchanges with a public generative AI platform were not protected by attorney-client privilege. The reason was simple: The platform was not the executive’s lawyer, its data practices did not support a reasonable expectation of confidentiality and the communications were not made to get legal advice from counsel.

In a Delaware case this year, a CEO’s interactions with a chatbot proved to be crucial in a ruling against him. The court quoted from the CEO’s AI chatbot conversations where he asked for help devising a strategy to avoid making a potential $250 million earnout payment. The court used those exchanges as evidence of the CEO’s motive, and also noted that some AI logs had been deleted.

The takeaway: AI interactions are not off-the-record brainstorming, and the same is true of AI-generated draft minutes. They will likely become part of a case and a court may focus on them if it thinks they reflect a director’s or executive’s most candid thoughts.

A communication between a director and an AI tool does not become privileged just because the topic is legal. In one case, a CEO’s use of a chatbot to strategize about how to avoid making an earn-out payment was cited by a court in a ruling against the CEO.

It’s Unclear if Work Product Protection Is Available

The work product doctrine protects materials prepared in anticipation of litigation from being disclosed to an adversary. Courts are still deciding how that doctrine applies to AI-assisted materials, and the early answers are inconsistent.

In the New York case above, the court concluded that AI-created documents were not protected work product. Even if they were prepared in anticipation of litigation, they were not prepared by counsel or at counsel’s direction, as the doctrine requires.

A federal court ruling in Michigan reached a different result. It held that a pro se litigant’s AI-assisted drafting materials were protected as work product and the court held that using ChatGPT did not waive the protection, because waiver usually requires disclosure to an adversary or disclosure that is likely to put the material in an adversary’s hands. However, this case involved a litigant representing himself without lawyers.

We should point out that the work product doctrine may not apply to ordinary minute-taking. Even when the board meets to discuss pending or threatened litigation, work product protection may not apply if the minutes were not prepared by counsel or at counsel’s direction. (Under Delaware law, drafts of minutes prepared by lawyers are privileged.)

Litigation Holds Can Apply to AI-Generated Content

When litigation is anticipated, directors and other and others at the company have a duty to preserve records that could be relevant in the dispute. That includes AI-generated minutes, transcripts, summaries and other content, which should be preserved if litigation is clearly on the horizon. Routine deletion of AI transcripts may be defensible in normal circumstances and may even be good information hygiene. But once a duty to preserve applies, if records are deleted, that could open up a company and its board to accusations of spoliation of evidence. A company’s retention policy should address AI-generated materials and explain when preservation obligations override routine deletion.

Use AI Thoughtfully – Don’t Avoid It

None of this means boards should avoid AI tools. The technology can save meaningful time and the legal issues can be managed with the right approach. Boards should consult with counsel before adopting these tools and should put practical guardrails in place. Those guardrails may include:

Deciding whether or not to transcribe meetings or parts of meetings.
Using AI tools in ways that protect privilege and work product where possible.
Setting clear procedures for review, approval, retention and deletion of AI-generated records.

Directors should approach AI the same way they approach other governance tools: Ask the right questions, get counsel’s input and make informed choices. Used carefully, these tools can support and enhance director’s work.

View other articles from this issue of The Informed Board

See all the editions of The Informed Board

This memorandum is provided by Skadden, Arps, Slate, Meagher & Flom LLP and its affiliates for educational and informational purposes only and is not intended and should not be construed as legal advice. This memorandum is considered advertising under applicable state laws.