• AI meeting tools can streamline the preparation of board minutes but may also create a parallel record that later becomes evidence in litigation. A recent Delaware Court of Chancery decision considered differences between AI-generated transcripts and official minutes in assessing a board’s objectives and decision-making process. Our alert discusses practical controls for boards and management teams considering whether and how to use AI meeting tools.

The Delaware Court of Chancery recently issued a decision that serves as a reminder that boards should establish clear controls before using AI meeting tools because AI-generated recordings, transcripts, and summaries may later become evidence used in litigation.

On August 28, 2026, Vice Chancellor Will issued a post-trial opinion in ATG Capital Opportunities Fund LP v. Lane, holding that the board of Empery Digital, Inc. (“Empery” or the “Company”) improperly rejected an activist investor’s director nomination notice and in so doing breached its fiduciary duties. In reaching its decision, the Court considered not only official board minutes but also AI-generated transcripts from the same board meetings. The two records of the same meetings differed, and the Court relied on the AI-generated transcripts to supply additional context that was not present in the official minutes about the board’s objectives and decision-making process.

Court’s Discussion of AI-Generated Transcripts. While its opinion was not focused on the use of AI, the Court cited AI-generated transcripts of board deliberations when analyzing two key issues.

First, after plaintiff ATG accumulated more than 10% of Empery’s stock and signaled a proxy contest, Empery’s board adopted a stockholder rights plan (“Rights Plan”), which would be triggered if a stockholder acquired 12.5% or more of Empery’s stock. The official minutes from the meeting at which the board adopted the Rights Plan reflected that the plan was intended to protect stockholders while Empery “continu[ed] to execute on its plan to close the NAV gap.” The AI-generated transcript, however, also included the Co-Chief Executive Officer telling the board that capping ATG’s ownership at 12.5% would limit ATG’s financial upside, making it “not worth it to put up [a] slate to go through the proxy solicitation process.” The transcript also reflected the same executive advising that the Rights Plan was “necessary in order for the board to remain in its position.”

Second, the board rejected ATG’s nomination notice on three primary grounds. The first ground was that ATG had not disclosed another stockholder as a participant in ATG’s solicitation. In assessing this assertion, the Court cited the AI-generated transcript of another Empery board meeting at which the Co-CEO expressed uncertainty about whether ATG and the other stockholder were acting together. According to the transcript, the executive said that ATG and the other stockholder “may still be acting[] as a group together”—although he did not “know it to be true”—and that this “narrative” was a “powerful part of th[e] rejection.”

The Court ultimately held that Empery’s bylaws did not require the disclosures Empery’s board demanded from ATG in rejecting ATG’s nomination notice, so the board’s decision to exclude ATG’s nominees lacked a contractual basis and, under Coster v. UIP Companies, Inc., 300 A.3d 656 (Del. 2023), was unreasonable in response to the threat perceived by the board. Although the AI-generated records did not decide the case, they formed part of the contemporaneous record the Court used to assess the board’s deliberations and motivations.

The potential evidentiary risk from use of AI tools extends beyond AI-generated meeting transcripts. For example, Fortis Advisors, LLC v. Krafton, Inc., 354 A.3d 906 (Del. Ch. Mar. 16, 2026) involved a breach-of-contract claim arising from Krafton’s termination of key executives and seizure of operational control after acquiring a video-game studio. The trial record included exchanges that Krafton’s CEO had with ChatGPT about how Krafton could avoid its contractual obligations. The Court considered those exchanges, together with other contemporaneous evidence, in finding that Krafton’s stated reasons for the terminations were pretextual.

Key Considerations. Boards considering the use of AI tools, including to record, transcribe, or summarize board or committee meetings should consider the following, in addition to the factors set forth in our prior client alert.

• Determine Whether a Recording Should Be Created. As a threshold matter, boards should decide with counsel and the corporate secretary whether to use AI meeting tools and, if so, in what circumstances. An overriding concern is the potential for verbatim transcripts to impede the candor of certain discussions that are important for the board in fulfilling its oversight role. In addition, recordings, transcripts, and summaries can create a near-verbatim “shadow record” that may be inaccurate, misleading, incomplete, or contain informal side comments without proper context. To the extent that boards find AI meeting tools helpful for certain purposes (such as the generation of accurate minutes), boards should weigh the perceived benefits of using such tools against potential risks from creating or retaining AI-generated materials.

• Create Clear Guidelines. The board should agree on clear, easy-to-follow guidelines governing the use of AI meeting tools and consider periodic updates for the board on emerging technologies and uses. Guidelines help promote consistent use and reduce the risk of creating or retaining unintended AI-generated board materials.

• Use Approved Tool and Vendor Controls. Use only tools specifically approved for board use after review of the product, configuration, and contract. Confirm that recordings and related data will remain confidential, will not be used to train models, will not be accessible to vendor personnel or shared with third parties except in specifically agreed circumstances, and will be protected by appropriate cybersecurity, access, storage, and deletion controls.

• Protect Privilege and Confidentiality. Even if the board permits an AI recording tool in some circumstances, it should consider prohibiting or restricting its use in meetings involving legal advice, litigation, strategic transactions, regulatory matters, material nonpublic information, trade secrets, or other highly confidential or sensitive topics.

• Obtain Notice and Consent. Determine which recording and consent laws apply, especially when participants join from different jurisdictions. Participants should receive clear notice, understand what the tool will create, and have a practical opportunity to raise concerns before discussion begins.

• Control Access and Circulation. Define who may activate an AI meeting tool, access its outputs, and circulate the resulting materials. Disable automatic dissemination and ensure that access and circulation comply with the board’s policy and applicable information barriers.

• Address Retention, Discovery, and Litigation Holds. Identify each record an AI tool creates (including recordings, transcripts, prompts, summaries, and drafts), where each such record is stored, and whether the vendor retains copies. Set a documented retention period in the ordinary course and apply it consistently, but suspend deletion whenever an applicable litigation hold or other preservation duty applies.

• Consider Similar Controls for Management Meetings. Companies should also consider the aforementioned issues when using AI tools to record, transcribe, or summarize management meetings, as such meetings implicate many of the same risks raised in the board meeting context.

This publication is for general information purposes only. It is not intended to provide, nor is it to be used as, a substitute for legal advice. In some jurisdictions it may be considered attorney advertising.