Krafton, Subnautica 2, and the ChatGPT Courtroom Battle

The ongoing dispute between South Korean publisher Krafton and the ousted founders of the Subnautica 2 studio Unknown Worlds has become one of the most unusual legal sagas in modern games. What began as a business conflict over a massive earnout has evolved into a case that now scrutinizes how a CEO allegedly used ChatGPT to shape a takeover strategy—and whether key employees tampered with corporate AI accounts and data on their way out.

This article breaks down the broader conflict, explains why ChatGPT logs have become a pivotal issue, and explores what this episode reveals about AI, corporate governance, and the future of game development deals.

Background: How Krafton Came to Own Subnautica

Unknown Worlds Entertainment rose to prominence with Subnautica, an underwater survival game that won critical praise and built a passionate community. Wanting to secure that creative talent, Krafton acquired Unknown Worlds in 2021. According to the Delaware Court of Chancery’s opinion on the case, the transaction included:

• $500 million upfront for the acquisition of the studio.
• Up to $250 million in contingent “earnout” payments, tied to the success of future titles such as Subnautica 2.¹
• Contractual promises that the founders would retain operational control and could only be removed for cause.¹

An earnout is a common feature in mergers and acquisitions, especially when buying creative businesses. The buyer pays a large sum upfront, then agrees to additional payments if specific performance targets are hit later. A 2022 analysis of M&A trends by Deloitte notes that earnouts are frequently used when there is uncertainty about future performance but strong confidence in leadership or IP.²

That structure put enormous financial weight on how Subnautica 2 would be handled—and on the relationship between Krafton and the original founders.

The Relationship Breaks Down

As development on Subnautica 2 advanced, the relationship between Krafton and key Unknown Worlds leaders—cofounders Charlie Cleveland and Max McGuire and CEO Ted Gill—began to deteriorate. Public court documents and coverage describe a clash on several fronts:

• Scheduling and release strategy: The founders wanted to proceed with an early access launch on a timeline they felt was right for the studio and community. Krafton became increasingly concerned about pacing, quality, and oversight.
• Control of the studio: The founders argue that Krafton gradually tightened its grip, restricting operational independence that had been part of the acquisition’s core promises.
• Earnout stakes: Both sides knew that a successful launch could trigger a large earnout payment. Founders claim Krafton wanted to avoid or minimize this obligation; Krafton counters that it simply sought to protect the game and company value.

The tension culminated when Krafton took significant governance steps, including altering board composition and delaying release plans. The Delaware court later found that prior to Krafton’s intervention, Subnautica 2 was on track for its early access release under Gill’s leadership.¹

Terminations, Alleged Data Theft, and the Start of Litigation

In mid-2025, Krafton dismissed Cleveland, McGuire, and Gill. The terminations triggered a flood of accusations on both sides and set the stage for the current legal fight.

Krafton’s Allegations

In court filings and responses, Krafton accuses the removed leaders of serious misconduct, including:

• Massive data downloads: The publisher claims the founders executed large-scale downloads of company files related to Subnautica 2 as their relationship with Krafton worsened.
• Theft of confidential information: Krafton characterizes these downloads as theft of trade secrets and proprietary data, supposedly to support plans to self-publish the game or gain leverage.
• Disloyal conduct: The company alleges the founders neglected duties, pursuing other interests and undermining Krafton’s goals.

The Delaware Court of Chancery dealt directly with some of these claims. It concluded that the data downloads did not meet the contractual threshold of “intentional dishonesty” and that the record did not credibly show the founders were stealing information to launch a competing product.³

The Founders’ Claims

The ousted leaders paint a very different picture. In their narrative, Krafton:

• Planned a takeover strategy to wrest control of Unknown Worlds while avoiding the earnout payout.
• Used legal and operational pressure to sideline the founders and stall Subnautica 2.
• Broke core promises of the acquisition agreement, especially around operational autonomy and protections against termination without cause.

Central to their argument is the claim that Krafton’s leadership, especially CEO Kim Chang-han, became fixated on minimizing or escaping the earnout. That fixation, they say, led to a strategy that weaponized the company’s power over publishing and internal systems.

Enter ChatGPT: AI Advice in a High-Stakes Corporate Conflict

The twist that pushed this lawsuit into global headlines is the allegation that Krafton’s CEO turned to ChatGPT, OpenAI’s conversational system, for advice on how to handle the dispute with Unknown Worlds and its founders.

According to the Delaware court’s opinion and subsequent reporting, the founder side argues that Kim asked ChatGPT to help brainstorm ways to manage or avoid earnout obligations and formulate a takeover strategy. The AI’s responses, they say, outlined a pressure-and-leverage roadmap that Krafton largely followed.^1,4

What ChatGPT Allegedly Advised

Court documents and legal commentary describe the alleged AI-generated strategy roughly along these lines:

• Apply coordinated pressure: Combine legal, financial, and operational tactics to push the founders into conceding on earnout terms.
• Secure “control points”: Lock down distribution and publishing channels—such as Steam permissions—to ensure Unknown Worlds could not release Subnautica 2 independently.
• Prepare for litigation: Develop a robust legal defense, compile internal memorandums, and build a narrative that justified stronger corporate intervention if negotiations failed.

Legal analysis by law firm DarrowEverett notes that Krafton did, in practice, implement many of these measures, especially around controlling Steam publishing rights and formalizing legal positions about early access, earnout disputes, and leadership removal.³

Using AI tools for brainstorming or drafting is increasingly common across industries. A 2023 survey by McKinsey reported that roughly one-third of respondents had used generative AI regularly for work-related tasks, with usage highest in tech and knowledge-intensive sectors.⁵ The unusual part here is not that ChatGPT was used, but that the outputs allegedly informed a contentious corporate strategy and then became evidence in litigation.

ChatGPT Logs Become Evidence

The role of ChatGPT shifted from background tool to central battleground when the founders sought detailed records of how Krafton’s executives had used the chatbot. In court, they argued that these logs could show:

• Intent: Whether Krafton’s leadership deliberately sought ways to evade contractual obligations.
• Planning: How closely actual corporate actions mirrored the AI-generated recommendations.
• Awareness: Whether executives knew they were pushing against the spirit or letter of the deal.

Krafton resisted expansive discovery of ChatGPT logs, arguing that the request was overbroad, burdensome, and a fishing expedition. The company describes the push for chatbot records as a last-ditch effort to find something incriminating after other claims faltered.

Accusations of Spoliation and “Incriminating” Data

Spoliation—the destruction or alteration of evidence—is a serious charge in civil litigation. Krafton now alleges that the ousted leaders themselves may have tampered with or deleted data, including AI account histories, in anticipation of legal action.

In filings described in news coverage, Krafton claims that on the day they were terminated, one of the founders told the others in a group chat to check corporate ChatGPT accounts and ensure there was “nothing incriminating” in them. The publisher argues that this message shows an intent to scrub potentially damaging information from company systems.

The founders, in turn, frame this as a benign or misunderstood comment, and continue to push for disclosure of the CEO’s ChatGPT usage, which they see as central to proving Krafton’s bad faith in managing the earnout and takeover.

Key Issues at the Heart of the Dispute

Aside from the headline-grabbing AI angle, the case raises several enduring questions about game industry acquisitions and corporate governance. These can be grouped into a few themes:

Why This Case Matters for the Wider Industry

While the dispute is specific to Subnautica 2, the implications extend far beyond one studio and one publisher.

1. Earnouts and Creative Autonomy

The case highlights how earnouts can create tension between creators and acquirers. When large contingent payments hinge on performance, both sides are incentivized to interpret contracts aggressively. The Subnautica conflict underscores the need for:

• Clear governance clauses that spell out decision-making power over release timing, marketing, and key milestones.
• Transparent dispute resolution mechanisms to handle disagreements before they escalate into full-blown litigation.
• Balanced incentives that reward success without encouraging either party to game the system.

2. AI Tools as Corporate Actors in the Background

The visible role of ChatGPT here is a preview of a larger trend: AI systems shaping internal strategy, legal drafts, and negotiations. That raises practical questions:

• How should companies archive AI interactions that influence major decisions?
• Who owns chats generated through corporate accounts—the individual, or the employer?
• When do AI logs become discoverable evidence in lawsuits?

Regulators and courts are only beginning to grapple with these issues. The EU’s AI Act, for instance, emphasizes transparency and risk management for high-impact AI systems, but corporate use of general-purpose tools like ChatGPT for strategy is still largely governed by general data and governance rules rather than AI-specific law.⁶

3. Data Governance and Exit Scenarios

The dispute over data downloads and alleged spoliation also highlights the importance of robust data governance. For studios and publishers alike, this case underlines the need to:

• Define clear policies for source code, documentation, and design assets, especially in the context of acquisitions.
• Manage access and logging so that both legitimate backups and suspicious downloads can be distinguished and audited.
• Specify treatment of accounts (including AI tools) upon termination or leadership transitions.

Likely Outcomes and What to Watch Next

The litigation between Krafton and the Unknown Worlds founders is complex, with multiple phases and claims. The Delaware Court of Chancery has already issued findings on some issues, including a determination that Krafton breached certain core commitments and that the founders’ data downloads did not constitute the kind of dishonesty that would justify withholding earnout protections.^1,3

That said, the fight over ChatGPT logs and alleged spoliation continues to shape the next stage of the dispute. Key questions that will influence the final outcome include:

• How much of Krafton’s internal AI usage must be disclosed, and in what form?
• Whether any missing or altered logs can be proven—and if so, whether courts will impose sanctions or evidentiary inferences.
• How damages and remedies will be calculated if the founders prove Krafton intentionally undermined the earnout’s spirit.

Regardless of the ultimate legal outcome, the case will likely be cited for years as an early example of generative AI’s entanglement with high-stakes corporate decision-making.

Practical Lessons for Game Studios and Publishers

For other developers and publishers contemplating acquisitions or large investment deals, the Subnautica 2 saga offers several practical takeaways:

• Negotiate AI and data practices up front. Contracts increasingly need language about data access, AI tools, and ownership of digital workspaces.
• Align expectations around release control. Creative autonomy is often a selling point in acquisitions, but it must be backed by enforceable governance terms.
• Document intent and decisions. When earnouts are large, documenting the business rationale for key decisions can protect both sides if disputes arise.
• Train leadership on AI risk. Executives using generative AI for sensitive strategic questions should understand how those interactions might later appear in court.

FAQs

What is the core dispute between Krafton and the Subnautica 2 founders?

The core dispute centers on whether Krafton breached promises made in the acquisition of Unknown Worlds—especially around operational control and protection from termination without cause—and whether its actions were designed to avoid paying up to $250 million in earnout payments tied to Subnautica 2.

Why are ChatGPT logs important in this case?

The founders argue that Krafton’s CEO used ChatGPT to help plan a strategy for taking control of the studio and minimizing earnout obligations. They believe the chatbot’s logs will show intent and planning that support their claims. Krafton portrays the request for logs as overly broad and suggests that the AI usage has been exaggerated.

Did the court say the founders stole data?

No. While Krafton alleged that the founders stole confidential information by downloading large amounts of data, the Delaware Court of Chancery found that these actions did not constitute “intentional dishonesty” under the contract and accepted that the data was kept confidential and then returned.³

Is using ChatGPT for corporate strategy illegal?

Using a tool like ChatGPT to brainstorm or draft strategic options is not inherently illegal. The legal issues arise from what decisions a company makes, how those decisions align with contracts and fiduciary duties, and whether relevant AI interactions are properly preserved and disclosed when they become evidence.

What does this mean for future game studio acquisitions?

The case underscores the need for precise contract drafting around earnouts, operational control, and data governance. It also signals that courts and counterparties may scrutinize AI usage in shaping strategy, so studios and publishers will need more explicit policies and documentation around how such tools are used.