A Delaware Court of Chancery opinion from August 2026 offers a valuable lesson about using AI tools to transcribe or summarize board meetings: those notes could end up in evidence.

In ATG Capital Opportunities Fund LP v. Lane, a proxy fight arose over Empery Digital, Inc., a Bitcoin treasury company. An activist hedge fund accumulated a significant stake and submitted a full slate of director nominees. The board rejected the nomination, however, claiming that the activist had secretly coordinated with a separate large stockholder and failed to disclose a substantial short position in Bitcoin ETFs.

Empery's advance notice bylaws did not in fact require the disclosures the board cited as grounds for rejection. Since the rejection was grounded in obligations that the bylaws did not impose, the court ruled that it was both contractually improper and inequitable under Delaware's framework for actions affecting the stockholder franchise.

The ‘smoking gun’ stemmed from AI-generated transcripts of the board meetings. Those transcripts were introduced as trial exhibits and quoted in the court's opinion, telling a different story than the board's formal rationale. While the board's official rejection letter cited specific bylaw deficiencies, the AI-generated meeting notes captured board members privately acknowledging that no single deficiency was actually detrimental on its own, and that the rejection letter simply “threw [in] the kitchen sink.” Another AI transcript revealed that the board chair stated the defensive rights plan was necessary to keep the board in place, an assertion absent from the formal meeting minutes. The court cited this material directly in finding that the board's stated rationale was pretextual.

The divergence from traditional board minutes is critical. This court's decision illustrates how AI meeting tools are changing the litigation landscape. As these tools become standard, records of real-time unfiltered discussions are likely to surface, and courts are paying attention.

Practical Considerations:

• Establish a clear policy governing AI meeting tools. Identify every AI transcription or summarization tool in use across board and committee meetings, and determine what records are being created, where they are stored, and who has access.
• Implement a review and approval process for AI-generated meeting records. Companies should build in steps to review AI-generated output from sensitive board meetings before it is retained or distributed.
• Treat AI-generated records as presumptively discoverable. Once created, these records are subject to litigation holds and discovery requests. Boards should factor that into how they structure deliberations on legally sensitive decisions.
• Review your governing documents to expressly address the requirements you intend to enforce. Courts will not read requirements into bylaws that the plain text does not contain. Periodic review with counsel ensures your documents reflect your actual intentions and will hold up when challenged.
• Evaluate consequential decisions with litigation in mind from the start. Counsel should be engaged at every stage, decisions should be grounded in valid legal authority, and rationale should be documented with care. Building a coherent, well-supported record from the outset positions you far better if that decision is later tested in court.