22 March 2026 · LinkedIn
"Fearing he had agreed to a 'pushover' contract, Krafton's CEO consulted an artificial intelligence chatbot to contrive a corporate takeover strategy."
That is a direct quote from a Delaware court ruling issued this week. Not a parody nor a leak: A judge's finding of fact, entered into the public record of one of the most consequential commercial courts in the world.
Changhan Kim, CEO of South Korean gaming publisher Krafton, wanted out of a $250 million contractual payout. His own legal team told him there was no clean route. His head of corporate development warned him in writing that firing the studio's founders without cause would mean "lawsuit and reputation risk." He had every institutional resource a major company can provide.
He ignored all of it, opened ChatGPT, and kept prompting until it gave him what he wanted. When the model told him the earnout would be "difficult to cancel," he pushed harder. It eventually produced a multi-stage takeover plan - codenamed "Project X" - complete with a messaging strategy for fans, which ChatGPT also drafted. Krafton followed most of it. The developers were fired. The studio was seized. The fan message backfired immediately.
The court ruled against Krafton on every count. The fired CEO is reinstated. The earnout window has been extended. The manufactured justifications were ruled pretextual. The comedy of it would be easier to enjoy if it weren't also a symptom.
We are entering a period in which AI will reshape commerce, governance, labour, and the rules of competition - faster than most institutions can track. The quality of leadership in that period is not an abstract concern. It determines, concretely, which futures remain available and which close off.
Kim did not fail because he used AI. He failed because he used it to avoid thinking - to launder a decision he had already made past the professionals whose entire function was to stop him making it.
The boardroom version of that failure is recoverable. A court ruling, a reinstated CEO, a nine-figure liability still pending.
The governance version is not.
The conversation about where AI takes us is already happening. The only variable is whether the people shaping it understand what they are holding - or whether they are still looking for a chatbot to tell them what they want to hear.
Join the conversation on LinkedIn →
← All writing