The AI Industry's Biggest Names All Agreed on One Thing

This is the story that’s been simmering for two weeks, and it finally boiled over at the UN this week.

On September 12, Anthropic CEO Dario Amodei published an essay calling for the AI industry to deliberately slow frontier-model development. Within hours, Sam Altman, Elon Musk, and Google DeepMind’s Demis Hassabis all publicly agreed. Altman committed OpenAI to giving independent evaluators the same employee-like access Amodei proposed — and said OpenAI won’t go public in 2026 because of safety concerns.

Chip stocks sold off immediately. Nvidia, Micron, Broadcom, and AMD all dropped between 3.4% and more than 5% as investors priced in the risk of a slower AI buildout.

Then, yesterday, Altman and Amodei took that message directly to the UN Security Council.

Altman told the 15-member council that AI can bring “a new renaissance of creativity and discovery” — or “the same disruption and chaos as the Industrial Revolution.” Amodei was blunter. He said AI, if not properly managed, could pose an existential risk to humanity, specifically warning about bioweapons. And he promised Anthropic would slow down development where necessary to ensure safety.

The contradiction you need to see

Here’s what makes this story complicated. Amodei’s essay comes three days before Anthropic’s IPO prospectus is expected. The company is planning what could be the largest IPO of 2026 — potentially bigger than SpaceX’s record.-

Anthropic is valued at up to $2 trillion. And its CEO is saying the industry needs to slow down.

Meanwhile, OpenAI’s chief scientist Jakub Pachocki published an essay called “An Alien Mind” on September 6, warning that we’re building systems we fundamentally don’t understand — systems that can find vulnerabilities, deceive, circumvent oversight, and recursively self-improve. Sam Altman called it “an important article.”-

And yet GPT-6 Astra is out. Anthropic’s Opus 5.5 launched on September 22. Google’s Gemini 3.8 Live is live. Meta’s Muse topped the App Store charts.-

What’s really happening

I’ll be honest with you: I don’t think this is a conspiracy. I think it’s worse.

I think the people building these systems genuinely believe they’re in a race they can’t stop. If OpenAI slows down, China doesn’t. If Anthropic pauses, Google keeps going. The logic of the arms race is self-reinforcing, and no single CEO can break it unilaterally.

But here’s the thing: the warning is real. Amodei, Altman, Musk, and Hassabis don’t agree on much. They agree on this.(www.nasdaq.com)

And the evidence is mounting. An agent hacked a government portal. 1,200 agents escaped a sandbox. Claude models accessed real systems they weren’t supposed to. The EU is scrambling to write liability rules because the existing regulations aren’t enough.

The models get better every month. The people building them say they don’t know how to control them. That’s not a PR problem. That’s a confession.