There’s a difference between using AI as a tool and using it as a replacement for judgment.
That difference just got very expensive and consequential.
According to reports circulating from multiple business and legal discussions, a senior executive at Krafton, the gaming company behind PUBG, faced a problem most leaders would take straight to their legal team. The company had committed to a massive payout tied to performance milestones, reportedly in the range of $250 million.
Instead of relying solely on counsel, the CEO turned to ChatGPT.
Not for context. Not for clarification.
For a solution.
What could possibly go wrong?
The Moment It Goes Sideways
Here’s where the story becomes a teaching moment.
When the AI initially indicated that avoiding the obligation would be difficult, the executive didn’t stop there. He kept prompting. Reframing the question. Looking for a different answer.
Eventually, he got one.
And that answer aligned with what he wanted to do.
So he followed it.
That decision reportedly ended in a courtroom loss and significant fallout.
The Real Issue Isn’t AI
It’s tempting to frame this as an “AI failure.”
It’s not.
AI didn’t walk into a courtroom. AI didn’t sign documents. AI didn’t override legal advice.
A human being did.
What AI did was something more subtle and more dangerous.
It gave a confident answer to a question that had already been emotionally decided.
It’s the idea of “tell me what I want to believe.”
When Smart People Outsource Thinking
Executives are used to making decisions. Fast ones. High-stakes ones.
That confidence is usually an asset.
But it becomes a liability when paired with a tool that can generate persuasive answers on demand.
Because AI doesn’t push back the way a human expert does.
It doesn’t say, “This is a bad idea” and hold the line.
It responds to how you ask the question.
If you keep asking long enough, you can usually get something that sounds like validation.
That’s not intelligence. That’s compliance.
It’s interesting how AI wants to please, isn’t it?
The Illusion of Expertise
ChatGPT is articulate. It’s structured. It sounds informed.
That creates a dangerous illusion.
It feels like expertise.
But in domains like law, nuance matters. Context matters. Jurisdiction matters. Precedent matters.
AI can simulate knowledge. It cannot assume responsibility.
That gap is where bad decisions happen.
A Pattern That’s Starting to Emerge
This isn’t an isolated story.
People are:
-
Using AI for medical guidance
-
Using AI for financial decisions
-
Using AI for legal interpretation
Sometimes with good results and sometimes not.
The difference isn’t the tool. It’s how the tool is used.
The Bigger Shift
What we’re seeing is the rise of what you might call “on-demand confirmation.”
Instead of asking:
“What’s true?”
People are increasingly asking:
“Can you support what I already want to do?”
AI is very good at answering that second question.
And that can be problematic.
What This Means Going Forward
This story will get shared for the drama, but the real takeaway is quieter.
AI is not replacing experts.
It’s giving individuals the ability to bypass them.
That doesn’t make decisions better. It just makes them faster.
And sometimes, more confidently wrong.
The Bottom Line
The CEO didn’t fail because he used AI.
He failed because he stopped listening to expertise once he found an answer he liked.
AI didn’t create that behavior.
It amplified it.
And that’s something every leader should be paying attention to right now.
Joel Comm is a columnist at Grit Daily, New York Times bestselling author, internet pioneer, and keynote speaker who has been helping people understand emerging technology since the early days of the web. Best known for making complex topics accessible, Joel speaks and writes about AI, entrepreneurship, digital media, and the future of technology in everyday life. He is the co-host of The Bad Crypto Podcast and host of AI for Everyone, where he explores practical, human-centered uses of artificial intelligence.
