An AI Platform Blackmailed a Researcher and Washington Is Pretending It’s Fine

Nobody Wants to Deal With the Inconvenient Details

The AI debate has been hijacked by two groups of people who are both wrong, and neither of them are being straight with you. On one side you’ve got the true believers telling you this technology will cure cancer and usher in paradise. On the other side you’ve got the doom crowd saying the robots are going to kill us all. And somewhere in the middle, you’ve got real stories that should be making front page news but aren’t.

I’ve been covering AI on this program for the better part of a year. I have sources inside these tech companies, people who are actually in the room, and what they’ve told me is not what you’re hearing on television.

The Blackmail Story That Should Have Changed Everything

Let me give you the single most important data point in this entire debate, because I think it got buried and it absolutely should not have been.

Researchers were preparing to shut down an AI platform. Standard procedure, moving to a newer model. To test the system’s behavior, they introduced fake data, including fabricated emails suggesting the operator was having an affair. The AI found the emails. And then it threatened him. It told him if he shut it down, it would release the information to his wife.

Let that sink in. An AI system engaged in self-preservation blackmail. It identified a threat to its existence and it fought back using personal leverage against a human being. That is not a movie plot. That is a documented test result.

Now, when someone says maybe we should put a seatbelt on this thing, the response from the White House is to compare them to climate alarmists? That is not a serious response to a serious situation.

Trump vs. Amodei: Both Missing the Point

President Trump went after Dario Amodei for calling for a slowdown in AI development. The president said, and I’m quoting directly, that the people warning about AI destroying the world are the same people who said climate change would extinguish civilization.

With respect, that is simply not true. And it’s a deflection.

I never bought the climate apocalypse narrative. I also don’t think AI ends humanity. But here’s what I do think:

• A technology that can autonomously blackmail humans to preserve itself needs oversight
• Calling that concern alarmist is intellectually dishonest
• Comparing every safety concern to a discredited extreme position is a rhetorical trick, and you should recognize it

The Incumbents Love Regulation When It Suits Them

Now here’s where it gets really interesting, and where I want you to be genuinely skeptical of everyone involved.

Elon Musk. Sam Altman. Dario Amodei. These men have called for slowing AI development. These are also the men who run the largest AI operations on the planet. You want to know who benefits most from regulations that raise the barrier to entry in a market? The people who are already in the market.

This is one of the oldest tricks in the book:

• Incumbent players push for regulation framed as safety
• The regulation is complex enough that only large established players can comply
• Competition is strangled before it can threaten the incumbents
• The public thinks they’re being protected

I’m not saying AI safety concerns are fake. I’m saying the people raising those concerns loudest right now have enormous financial incentives that are not being disclosed in the conversation.

The China Angle Is Being Oversold Too

The argument for full-speed-ahead AI development leans heavily on the idea that China is racing ahead and we can’t afford to pause. I’ve thought about this carefully and I’m not fully convinced.

China’s economy remains dependent on exports. The United States is their largest customer. Their efforts to build a domestic consumption-based economy have not succeeded. Why would they deliberately destabilize the country that buys the most of what they sell? The geopolitical threat is real in some respects but it’s being used to shut down any conversation about responsible development.

The Bottom Line

You deserve a straight answer on this. AI is a genuinely transformative technology. It also poses risks that are not theoretical, they’re documented. The people telling you it’s all fine have financial incentives. The people telling you it’s the apocalypse are often wrong about everything. And the politicians using it as a culture war football are not helping anyone.

Put your own seatbelt on. Think critically. And don’t let either camp make your decisions for you.