OpenAI’s Trust Problem Isn’t Going Away

OpenAI has a timing problem. Or maybe it’s just a Sam Altman problem.

On the same day the company published a policy paper about how it plans to keep humanity safe when superintelligence arrives, The New Yorker published a sprawling investigation into whether CEO Sam Altman can actually be trusted to follow through on those promises.

Reading both side by side is almost surreal.

OpenAI’s paper lays out the grand vision: push for policies that “keep people first” as AI begins “outperforming the smartest humans even when they are assisted by AI.” The company promises to stay “clear-eyed” about risks, including AI systems escaping human control or governments using AI to undermine democracy. Without proper mitigation, “people will be harmed.” The whole thing reads like a company that has its priorities straight.

Then you have The New Yorker piece, which paints a very different picture. The headline gives it away: the problem isn’t the technology, it’s the guy in charge.

I’ve been around long enough to know that these kinds of coincidences don’t happen by accident. Either someone at OpenAI genuinely thought this was the right week to talk about trust and safety, or they’re completely disconnected from how the rest of the world sees them.

The policy paper itself isn’t bad. It covers the right bases: monitoring for extreme scenarios, advocating for broad distribution of AI benefits, pushing for transparency. But none of that matters if the people running the show can’t be trusted to follow through when the pressure mounts.

And that’s the core issue. OpenAI has a long history of promising one thing and doing another. The non-profit structure was supposed to keep things safe, then it wasn’t. The safety team was supposed to have teeth, then it didn’t. The company was supposed to be transparent, then it stopped being transparent.

Altman himself has been at the center of multiple controversies. The boardroom drama. The safety researcher departures. The pivot from open to closed. Each time, the company says it’s learned its lesson. Each time, the same patterns emerge.

So when OpenAI releases a paper saying, “Trust us, we’ll handle superintelligence responsibly,” the natural reaction isn’t relief. It’s skepticism. And that skepticism is earned.

The New Yorker piece isn’t just a hit job. It’s a reflection of what people inside and outside OpenAI have been saying for years. The company’s biggest challenge isn’t technical. It’s credibility.

Superintelligence may or may not arrive in our lifetimes. But OpenAI’s trust deficit is here right now, and no amount of well-written policy papers is going to fix it.