Delaware Court Leverages AI Chat Logs to Decide $27M Bonus Dispute Against Krafton
A Delaware court has delivered a landmark ruling using Krafton CEO Kim Chang-han‘s conversations with the AI chatbot ChatGPT as key evidence in a $27 million performance bonus dispute. The judge found that Krafton’s internal moves to dismiss executives from an acquired studio were based on advice extracted from AI, exposing a new legal vulnerability for companies.
Vice Chief Judge Lori Will ruled that the dismissal was a deliberate tactic by Krafton to avoid paying about 3.7 billion Korean won (approximately $27 million) in bonuses tied to the success of the sequel to the survival game Subnautica. The case underscores that conversations with AI chatbots do not enjoy attorney-client privilege, making them fully accessible during litigation.
How AI Chat Logs Became the Smoking Gun
Krafton acquired Unknowm Worlds, developer of Subnautica, in 2021, committing up to $250 million in bonuses if the game sequel met sales targets. In May 2026, internal data projected a likely payment of $200 million. Within two months, Krafton abruptly fired the CEO and two executives from Unknowm Worlds, claiming they rushed the game’s release in an unfinished state.
The fired executives contested the dismissals, arguing they were a pretext to dodge bonus payments. The court sided with them, bolstered by Kim Chang-han‘s own ChatGPT interaction logs. When warned dismissal wouldn’t nullify bonus obligations, Kim turned to AI for a strategy. While ChatGPT initially urged caution, it later suggested creating an internal task force or forcibly renegotiating the deal. Krafton executed these AI-inspired steps under the codename “Project X”, then dismissed the executives.
Kim claimed during trial he treated ChatGPT like a search engine, not seeking legal advice. The court rejected this, highlighting how he shared chatbot conversations over Slack and email, which are subject to discovery. Suspicion deepened when Kim admitted deleting parts of his ChatGPT chats, citing privacy concerns, while other related records remained intact—suggesting selective deletion. This intensified the court’s rulings against Krafton.
AI Chats Lack Legal Protections Unlike Lawyer Talks
This ruling shines a new spotlight on a critical legal gap: AI chatbot discussions have no confidentiality guarantees under U.S. attorney-client privilege laws. Unlike private talks with lawyers, AI interactions can be discovered and introduced as evidence. This principle echoes a February ruling by the U.S. District Court in New York, which ordered submission of documents created with AI platform Anthropic’s Claude, declaring no attorney-client relationship can exist with AI.
Top law firms like New York’s ShutreMonte and Debevoise & Plimpton are issuing urgent client guidance. They warn that sharing sensitive or legal advice via AI can nullify privilege protections. One firm suggests including legal counsel directives in AI prompts to bolster privilege claims. Experts bluntly caution executives: if you use AI for aggressive corporate maneuvers or litigation strategies, expect opposing counsel to scrutinize every word.
What This Means for Executives and Companies Nationwide
As AI tools become standard for exploring business strategies, the Krafton case is a stark warning that executives must treat AI conversations as public records. Any sensitive discussions about litigation, restructuring, or corporate strategy should occur directly with legal counsel, not AI, to maintain confidentiality.
The Delaware ruling delivers a powerful message across South Carolina and the U.S.: AI’s speed and convenience come with risks. Executives relying on AI for critical decisions now face the real threat of those chats being weaponized in court. Krafton’s CEO’s ChatGPT logs became the decisive evidence proving intent—turning a new chapter in corporate law and AI use.
With AI widely embraced in business, companies must urgently rethink how they manage communications involving strategy and legal risks. Expect more courts to follow Delaware’s lead in demanding transparency from executive AI consultations, marking a profound shift in corporate governance and litigation.
Vice Chief Judge Lori Will: “No attorney-client privilege extends to conversations with AI chatbots, making these records fully discoverable.”
