Kobre & Kim's Special Situations Disputes Team
June 11, 2026
Managing AI Governance Risks as U.S. Courts Expand Scrutiny of Corporate Decision-Making
Recent U.S. court decisions, including a recent Delaware Court of Chancery ruling, suggest that AI prompts, chat logs, and outputs may be discoverable and used as evidence in corporate disputes, creating new governance and litigation risks for companies. As AI becomes increasingly embedded in business decision-making, organizations should ensure appropriate oversight, documentation, and retention practices to help manage potential exposure.
Artificial Intelligence (AI) is now embedded across business planning, forecasting, and post‑closing operations—often shaping strategic and operational decisions without clear governance, documentation, or visibility into how those tools influence outcomes. While many users still treat AI interactions as informal or transitory, those communications increasingly form part of the record in high-stakes corporate disputes.
Our prior client alert focused on privilege and confidentiality risks associated with AI use, particularly where employees or executives rely on third‑party tools. Recent decisions highlight a consequential next step. In United States v. Heppner, a federal court held that interactions with a publicly available AI platform—both prompts and outputs—may be discoverable and may not be afforded attorney-client privilege or work-product doctrine protections.
The Delaware Court of Chancery’s March 16, 2026 decision in Fortis Advisors, LLC v. Krafton, Inc. illustrates how such communications—once admitted into the record—may be deployed. The Krafton case arose from a post-closing earnout dispute in which the seller alleged that the buyer made operational and strategic changes—such as shifting resources, altering product priorities, and adjusting commercialization strategies—that effectively depressed the performance metrics used to calculate post-closing earnout payments. The buyer argued these actions fell within its contractual discretion and reflected legitimate post-acquisition integration decisions.
The court treated internal AI interactions as contemporaneous evidence of how operational strategies were created and decisions were justified within the organization. Rather than relying solely on after-the-fact testimony or formal board materials, the Court used AI prompts and outputs as a real-time window into how the buyer’s management assessed the implications of decisions on potential earnout payments, tested alternative courses of action, and articulated business rationales within the company—evidence that could either support or undermine the parties’ competing accounts of intent and good faith.
These cases signal that AI use is no longer just an exposure issue, but a process and governance risk that courts may scrutinize closely.
Preparing for AI-Focused Scrutiny
- Assume AI Communications Will Be Evidence. Courts are increasingly willing to treat AI prompts, chat logs, and outputs as comparable to emails and internal memoranda for evidentiary purposes. The combination of Heppner and Krafton underscores that AI interactions may not only be discoverable but also used affirmatively to establish knowledge, intent, motive, or pretext.
- Understand that Courts May Treat AI Prompts as Statements of Intent. Prompting an AI tool can be framed much like communicating a thought process to a coworker, but the prompt itself may provide a clearer window into strategic intent. In Krafton, AI exchanges featured prominently in the court’s evaluation of why certain actions were taken, not just what actions occurred.
- Assess AI Use Risks in Busted Deals or Post-Closing Earnout and Covenant Disputes. AI-generated records may carry particular significance in M&A litigation, particularly in busted deals where the motives of the withdrawing party will be scrutinized. Additionally, post-closing disputes, such as earnout disputes or covenant-heavy transactional or financing arrangements, where operational decisions can affect milestone achievement, and parties may later allege efforts to circumvent negotiated constraints. Parties should evaluate which AI-related records exist, assess the risks those records may pose, and prepare to seek discovery regarding the opposing party’s AI use and related decision-making processes.
- Be Prepared to Defend Process, Not Just Results. Delaware courts routinely focus on how business decisions were made. Where AI inputs inform strategic or financial actions, companies are better positioned when they can demonstrate human oversight, independent judgment, and a clear path from AI inputs to final decisions, rather than an apparent reliance on AI outputs alone.
- Craft AI Use and Retention Policies with Litigation Risk in Mind. Policies should define when AI tools may be used for sensitive operational or deal-related decisions, clarify review and accountability responsibilities, and align retention practices with anticipated litigation risk. Recent cases, including Krafton, suggest that selectively deleting AI records may increase exposure rather than reduce it, much like selectively deleting emails or other internal communications. Companies should review AI retention and auto-deletion policies to balance information storage burdens with litigation risk management.
As AI becomes embedded into the M&A process and post‑closing operations, Delaware courts are likely to assess its use through familiar fiduciary and process‑based standards—standards that frequently drive outcomes in earnout and governance litigation. Companies that integrate AI into disciplined decision‑making structures and treat AI interactions as part of the corporate record will be better positioned to defend their conduct when those decisions are later challenged.