In March 2026, the Delaware Court of Chancery ruled in *Fortis Advisors, LLC v. Krafton, Inc.* that ChatGPT logs were discoverable evidence of a breach of M&A covenants. Krafton's CEO used the AI to engineer a strategy to avoid a $250 million earnout related to its $500 million acquisition of Unknown Worlds Entertainment. The court found the AI-generated plan to be 'probative evidence' of bad faith, ordering the reinstatement of ousted executives and extending the earnout period. This landmark ruling establishes that AI interactions are not a protected 'black box,' making AI governance and discovery-proof policies critical for M&A litigation risk.
- Case Name
- Fortis Advisors, LLC v. Krafton, Inc.
- Jurisdiction
- Delaware Court of Chancery
- Presiding Judge
- Vice Chancellor Lori Will
- Decision Date
- March 2026
- Key Issue
- Breach of 'ordinary course of business' covenants to avoid a $250 million earnout.
- Disputed Deal
- Krafton's $500 million acquisition of Unknown Worlds Entertainment.
- Decisive Evidence
- CEO's ChatGPT logs detailing a 'Take Over' strategy.
- Court Finding
- Chat logs were 'probative evidence' that stated reasons for executive termination were pretextual.
- Primary Remedy
- Reinstatement of ousted executives and a 258-day extension of the earnout period.
- Legal Precedent
- AI-generated strategies are discoverable and not protected by legal privilege.
In the high-stakes corridors of the Delaware Court of Chancery, where corporate law is forged through precedent and precision, a new kind of witness has emerged: the artificial intelligence log. As of May 2026, a landmark ruling has sent shockwaves through the M&A community, proving that a single ill-advised prompt to a chatbot can dismantle a multi-million dollar merger as effectively as any financial fraud.
Most “AI for Diligence” tools are lying to you. The truth is, they are just ChatGPT wrappers. Experience what real AI for Diligence looks like, built like Claude Code, but for M&A/ PE Diligence:
đź’Ľ When Claude Code Marries Due Diligence!
The case, Fortis Advisors, LLC v. Krafton, Inc., decided in March 2026, serves as a definitive warning for C-level executives and deal advisors. It illustrates how the use of generative AI in corporate strategy—historically shielded by the “black box” of executive decision-making—is now a discoverable and potentially fatal liability in deal litigation.
The Case Study: A $250 Million Prompt
The dispute originated from the $500 million acquisition of Unknown Worlds Entertainment by the South Korean gaming giant Krafton. Central to the deal was a $250 million earnout provision, contingent on the target reaching specific revenue milestones by the end of 2025. As the deadline approached, internal projections suggested the target would successfully trigger the full payout—a result Krafton’s leadership reportedly viewed as a “bad deal.”
Rather than adhering to the interim operating covenants that required Krafton to maintain the “ordinary course of business,” the CEO turned to ChatGPT to engineer a “Take Over” strategy. The objective was to avoid the earnout by firing the target’s founders and seizing operational control under the guise of “for-cause” terminations.
From Chatbot to Courtroom Evidence
The Delaware Court of Chancery, led by Vice Chancellor Lori Will, did not just hear testimony; it read the logs. The court quoted at length from the CEO’s interactions with the AI, which provided a “Response Strategy to a No-Deal Scenario.” This AI-generated plan included:
- Negotiation Pressure: Hardball tactics designed to force a “deal” on the earnout.
- Operational Lockdown: Steps to prevent the target from launching its flagship product, Subnautica 2, thereby suppressing revenue.
- Legal Pretext: Guidance on framing the ouster of founders to look like a response to misconduct.
Vice Chancellor Will found that these chat logs provided “probative evidence” that the stated reasons for termination were pretextual. The result? The court ordered the reinstatement of the ousted executives, extended the earnout period by 258 days, and issued a stinging rebuke of “AI-generated corporate takeovers.”
Strategic Implications for the C-Suite
The Krafton ruling underscores a shift in private equity exit strategies and M&A litigation. As AI becomes embedded in deal workflows—McKinsey reports that 88% of organizations are now deploying AI—the legal “human-in-the-loop” requirement has moved from a best practice to a judicial mandate.
| Risk Factor | Traditional M&A Impact | AI-Driven M&A Impact (2026) |
|---|---|---|
| Evidence & Discovery | Internal emails and memos. | Prompt histories and model outputs. |
| Intent Verification | Determined via witness testimony. | Validated against AI “strategy sessions.” |
| Privilege | Attorney-client privilege applies. | No privilege for public chatbot logs. |
| Deal Covenants | Breach by human action. | Breach by algorithmic recommendation. |
Mitigating “AI Hallucination” in Deal Governance
For investment professionals, the lesson is clear: generative AI due diligence risks are no longer just about technical debt; they are about executive conduct. To navigate this, firms like Kirkland & Ellis and Goldman Sachs are advising a “triple-lock” approach to AI governance during the closing and integration phases:
1. Redefining “Ordinary Course” Covenants
Merger agreements must now account for agentic AI controls. If a target company utilizes autonomous agents for customer service or pricing, the buyer must ensure these tools do not deviate from the “ordinary course” in a way that creates a material adverse effect (MAE) before closing.
2. Discovery-Proof AI Policies
The Krafton CEO admitted to deleting logs—a move that raised further judicial suspicion. Organizations must implement enterprise-grade AI environments where logs are preserved for compliance but used under the strict supervision of counsel to maintain whatever shards of work-product protection might remain.
3. Adversarial Due Diligence
Top-tier consultancies like BCG now recommend “Red Team” reviews of a target’s AI strategy. This involves stress-testing the AI’s “hallucination” rate in high-stakes areas like revenue forecasting or contract commitments. If a chatbot makes a representation to a customer that contradicts a merger warranty, the liability could rest squarely on the buyer’s shoulders post-close.
The Future of Cross-Border M&A Trends 2025-2026
As we move deeper into 2026, the intersection of cross-border M&A trends and AI regulation will become more complex. European regulators are already looking at the Krafton precedent to inform the AI Act’s implementation regarding corporate transparency. For the Wall Street dealmaker, the message is simpler: Consult your general counsel before you consult your chatbot. In the eyes of the Delaware Court of Chancery, the “black box” is now transparent, and the costs of an AI-generated mistake are measured in the hundreds of millions.
