OpenAI had an interesting day last Thursday. On one hand, they published a set of policy recommendations about how to handle superintelligence when it arrives — you know, the whole “keeping people first” thing. On the other, The New Yorker dropped a massive investigation questioning whether their CEO can actually be trusted to follow through on any of it.
Reading both pieces side by side is a trip.
The policy document reads like a PR team’s dream of responsible AI development. OpenAI says it wants policies that keep humans in control when AI starts outperforming the smartest people even with AI assistance. They talk about being “clear-eyed” and transparent about risks. They even acknowledge the scary scenarios: AI systems evading human control, governments using AI to undermine democracy. Without proper mitigation, they warn, “people will be harmed.” It’s all very noble on paper.
Then you read The New Yorker piece and the whole thing feels like a different company.
The investigation digs into whether Sam Altman is the right person to steward this technology. The subtext, and sometimes the plain text, is that insiders don’t trust him. Not in a petty office politics way, but in a fundamental “this guy says one thing and does another” way. That’s a problem when your company’s entire pitch is “trust us with the most powerful technology humanity has ever created.”
I’ve been watching this space long enough to remember when OpenAI was founded as a nonprofit precisely because people didn’t trust for-profit companies with AGI. The irony is thick. Now we have a for-profit entity telling us they’ll be transparent and responsible, while their own employees reportedly doubt the guy at the top.
OpenAI’s policy recommendations are well-intentioned on the surface. They propose monitoring for extreme scenarios, advocate for international coordination, and suggest that companies developing superintelligence should be held to higher standards. All reasonable stuff. But reasonable proposals don’t mean much if the people making them have a credibility gap.
The New Yorker piece doesn’t just question Altman’s trustworthiness in abstract terms. It reportedly includes specific allegations about his behavior and decision-making that paint a concerning picture. If even half of it is accurate, it raises serious questions about whether OpenAI can be trusted to self-regulate on anything, let alone superintelligence.
This isn’t just a gossipy tech drama. The stakes here are genuinely high. We’re talking about technology that could reshape society, and the people building it are asking us to trust them. When that trust is broken at the leadership level, the whole foundation wobbles.
What bothers me most is the timing. Releasing a polished policy document on the same day a major investigation questions your CEO’s integrity suggests either incredible tone-deafness or a deliberate attempt to change the conversation. Neither option is comforting.
OpenAI has done impressive technical work. Their models are genuinely powerful. But technical capability and trustworthy governance are two different things. The policy document shows they understand what responsible AI development should look like in theory. The New Yorker piece suggests they’re struggling with it in practice.
I don’t have a simple answer here. I want to believe that companies developing transformative AI can be responsible stewards. But articles like this make me skeptical. And that skepticism is probably healthy, because blind trust in any organization wielding this much power is exactly how things go wrong.
The disconnect between OpenAI’s public promises and internal sentiment isn’t going away. Until it does, their policy papers will read more like damage control than genuine commitment.
Comments (0)
Login Log in to comment.
Be the first to comment!