ChatGPT logs become key evidence in Krafton contract fight

In Fortis Advisors, LLC v. Krafton, Inc., 354 A.3d 906 (Del. Ch. 2026), a Delaware court relied on a CEO’s ChatGPT queries and later actions to find motive and pretext in a breach of contract dispute. Krafton CEO CH Kim sought ways to avoid an earnout owed after the company acquired Unknown Worlds Entertainment, despite legal advice warning against that path. ChatGPT responses said the earnout was “difficult to cancel” but suggested pressure tactics that led to an internal “Project X” plan to renegotiate the payment or take over the studio.

Krafton later terminated key employees for cause, citing “intentional acts of dishonesty,” but the court found the stated reasons were pretextual and that Krafton breached the agreement by taking operational control. The court reinstated Gill as CEO, restored his authority and extended the earnout period by 258 days, while leaving monetary damages for a later phase.

The dispute underscores how AI prompts, outputs and copied chatbot content can become trial evidence, especially when shared through Slack or other internal systems. Recent rulings remain fact-specific on privilege and work-product protection, turning on platform terms, counsel involvement and whether outputs were disclosed. Companies are urged to update retention policies, litigation holds and AI governance programs so prompts, responses, drafts and tool settings are preserved when legal risk emerges.