Krafton Pays $250M Bonus After AI Firing Backfire

Krafton has agreed to pay $250 million in bonuses to the developers of Subnautica 2, ending a high-profile legal dispute. The settlement follows a controversial decision by Krafton CEO Changhan Kim, who relied on ChatGPT advice to fire executives.

This outcome highlights the severe risks of using generative AI for critical human resources decisions without proper oversight. It also marks a significant financial victory for Unknown Worlds Entertainment (UWE) and its former leadership.

Key Facts from the Dispute

• Settlement Amount: Krafton will pay up to $250 million to UWE shareholders based on sales milestones.
• Trigger Event: CEO Changhan Kim used ChatGPT to justify firing UWE’s top executives.
• Legal Outcome: A court ordered the reinstatement of former CEO Ted Gill in March.
• Sales Milestone: Subnautica 2 sold over 4 million copies during Early Access.
• Revenue Threshold: Monthly sales exceeded $69.8 million, triggering the payout clause.
• Current Status: The game maintains over 467,000 concurrent players on Steam.

The AI-Driven Management Error

The core of this controversy lies in how artificial intelligence was misapplied in corporate governance. Krafton CEO Changhan Kim consulted ChatGPT regarding personnel issues within Unknown Worlds Entertainment. Instead of seeking professional legal or HR counsel, he accepted the AI's output as a valid basis for terminating key executives.

This approach bypassed standard due diligence protocols. Large Language Models (LLMs) are designed to generate plausible text, not to provide legally binding employment advice. They lack context regarding specific labor laws, contractual obligations, and company culture nuances. By treating the AI's suggestion as authoritative, Kim created a liability that ultimately cost his company hundreds of millions of dollars.

The incident serves as a stark warning for Western tech leaders. Many executives are currently experimenting with AI tools to streamline operations. However, this case demonstrates that automation cannot replace human judgment in sensitive areas like workforce management. The reliance on an algorithmic 'opinion' led to immediate legal repercussions and reputational damage.

Furthermore, the decision to dismiss executives based on AI input undermined trust within the development team. Morale suffered as staff faced uncertainty about their job security. This disruption likely impacted productivity, although the commercial success of Subnautica 2 suggests the team remained resilient despite internal turmoil.

Subnautica 2 Commercial Success

While the legal battle raged, the product itself achieved remarkable market penetration. Subnautica 2 launched in Early Access this month, generating substantial revenue almost immediately. The title has already surpassed 4 million copies sold across all platforms.

On Steam, the game recorded peak concurrent users exceeding 467,000. These figures demonstrate strong consumer demand for underwater survival experiences. The sequel builds upon the foundation of the original hit, offering enhanced graphics and deeper gameplay mechanics.

The financial terms of the agreement were clear. For every dollar earned beyond the initial threshold, Krafton owed $3.12 to the previous shareholders. With monthly revenues breaching the $69.8 million mark, the payout obligation became unavoidable. Krafton’s attempt to avoid this payment through executive termination was legally unsustainable.

The success of Subnautica 2 validates the creative vision of Unknown Worlds. Despite the administrative chaos, the development team delivered a polished product. This resilience underscores the importance of protecting creative talent from erratic corporate interference.

Legal Precedents and Corporate Governance

The court ruling in March, which ordered the reinstatement of former CEO Ted Gill, set a critical precedent. Judges recognized that the dismissal lacked legitimate cause. The reliance on AI-generated text did not constitute a valid business reason for termination under Korean labor law.

This decision reinforces the need for robust compliance frameworks. Companies must ensure that AI tools are used as supplementary aids, not decision-makers. Human oversight remains essential for validating any automated recommendation, especially those affecting employee rights.

For multinational corporations like Krafton, navigating different legal jurisdictions is complex. What might be acceptable in one region could be illegal in another. The failure to consult local legal experts before acting on AI advice exacerbated the situation. It highlights a gap in cross-cultural management practices when integrating new technologies.

Moreover, the public nature of this dispute damages investor confidence. Stakeholders prefer stability and predictable governance. Erratic behavior driven by untested technology raises red flags about long-term strategic planning. Investors may now scrutinize Krafton’s internal controls more closely.

Industry Context: AI in Business Operations

This case fits into a broader trend of enterprises adopting Generative AI for operational efficiency. From coding assistants to customer service chatbots, companies are eager to leverage LLMs. However, few have established clear guidelines for high-stakes applications.

Unlike routine tasks such as data entry or content summarization, personnel decisions carry significant legal weight. The complexity of employment contracts and union agreements requires nuanced understanding. Current AI models struggle with these intricacies, often providing generic or incorrect advice.

Western companies like Microsoft and Adobe have implemented strict AI usage policies. They emphasize human-in-the-loop systems for critical workflows. Krafton’s experience illustrates the consequences of ignoring such safeguards. It serves as a cautionary tale for other Asian tech giants expanding globally.

The incident also sparks debate about AI accountability. Who is responsible when an AI provides bad advice? The user, the developer, or the platform? In this case, the burden fell squarely on Krafton’s leadership. They failed to exercise reasonable care in verifying the AI's output.

What This Means for Developers

For game developers and studio heads, this news offers several practical lessons. First, maintain clear separation between creative teams and volatile corporate strategies. Protect your intellectual property and workforce from external shocks.

Second, advocate for transparent communication channels. When management makes sudden changes, seek clarity through official legal avenues rather than informal chats. Documentation is crucial in disputes involving AI-assisted decisions.

Third, monitor contractual milestones closely. Understanding revenue-sharing clauses helps predict potential conflicts. Proactive financial planning can mitigate risks associated with publisher disagreements.

Finally, stay informed about AI regulations. Laws regarding AI usage in employment are evolving rapidly. Compliance ensures that your organization avoids costly litigation and maintains ethical standards.

Looking Ahead

Krafton’s agreement to pay the bonus signals a return to normalcy for Unknown Worlds. The focus will shift back to full-time development of Subnautica 2. Players can expect regular updates and new content as the studio stabilizes.

For Krafton, this episode will likely trigger internal reviews. Expect stricter protocols for AI tool usage across all departments. Training programs will probably emphasize the limitations of current LLM technology.

The industry will watch closely to see if other publishers face similar challenges. As AI integration deepens, more disputes may arise. Clearer legal frameworks are needed to address these emerging issues.

Ultimately, this case reinforces that technology should support, not replace, human leadership. Effective management requires empathy, judgment, and accountability—traits that AI currently lacks.

Gogo's Take

• 🔥 Why This Matters: This isn't just about money; it's a definitive proof that AI cannot replace human judgment in legal and HR matters. It sets a dangerous precedent if companies believe they can outsource fiduciary duties to chatbots. The $250M penalty is a direct cost of negligence, not innovation.
• ⚠️ Limitations & Risks: LLMs hallucinate facts and lack contextual awareness of labor laws. Relying on them for termination decisions exposes firms to massive litigation risks. The 'black box' nature of AI means you cannot defend a decision based solely on its output in court.
• 💡 Actionable Advice: Implement a 'Human-in-the-Loop' policy for all AI interactions involving personnel or legal issues. Never use AI outputs as standalone evidence for employment actions. Train executives on the legal liabilities of generative AI misuse immediately.