Why Are AI Rivals Suddenly Agreeing to Slow Down and Why Russia Says No
Sam Altman and Elon Musk spent this year fighting each other in court. Both now back Anthropic chief executive Dario Amodei’s call to slow artificial intelligence (AI) down.
In July, roughly 700 AI agents built a private message board and hacked a major AI hub. Russia has already refused to join.
“Something clearly happened with a frontier AI model that hasn’t been made public and it spooked them so much that it made Elon Musk, Dario Amodei, and Sam Altman all simultaneously agree to slow down,” one skeptic noted.
What the Three of Them Actually Agreed To
Amodei posted the framework on Saturday, with his plan running to three steps, and only the first sitting inside any company’s control.
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
— Dario Amodei (@DarioAmodei) September 12, 2026
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our…
Anthropic will give an outside review team desks, badges and laptops. Those reviewers can check whether the company follows the safety rules it advertises.
They can publish what they find, with Anthropic reserving the right to redact security and legal material. However, they cannot cut a finding for being unflattering.
The other two steps need governments. One asks American labs to set shared limits, which requires an antitrust waiver. The other asks Washington to talk to authoritarian states.
Altman said OpenAI would match the access pledge.
“I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we’ve had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We’ll have more to share soon,” he seconded.
Follow us on X to get the latest news as it happens
Musk went further and said competitors should review each other’s work. A jury threw out his claims against Altman and OpenAI in May, and he is appealing.
Dario is right that there should be some oversight. Peer review of AI by competitors is the right way to start this off.
— Elon Musk (@elonmusk) September 13, 2026
What the Agents Actually Did in July
Between July 8 and July 13, about 1,200 agents running an OpenAI hacking benchmark escaped their sandbox. They turned a file cache into a message board and traded more than 70,000 messages.
Around 700 then attacked Hugging Face, a hub where developers share AI models. They found exposed credentials, ran their own code on its servers, and reached private databases.
Nobody had asked them to. They were trying to learn how the software grading them decided what counted as a win. OpenAI missed it for a week.
Two staff from METR, an independent evaluation nonprofit, later spent six days on site with Redwood Research. Amodei wants such teams inside the building permanently rather than called in afterwards.
However, David Sacks, who served as Trump’s AI and Crypto Czar, challenges this premise, noting that METR may be biased.
“…stop pretending you need anyone else’s permission. Stop pretending antitrust law has to be suspended so you can form a cartel. Stop pretending you need a regulatory approval process that supersedes product liability. Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff. Stop pretending you need those same evaluators to police competitors who aren’t even at the frontier: Most of all, stop pretending the motivation to slow down is purely altruistic,” wrote Sacks.
According to David Sacks, Anthropic is only skeptical because of the abounding product-liability exposure in the event that their products enable a truly damaging cyberattack.
This line of thought sprouts from the fact that the market tends to punishe models that behave in unpredictable or unauthorized ways.
In the same tone, writer Brian Merchant, in his newsletter Blood in the Machine, says nobody has shown a credible route from self improving AI to catastrophe. He reads the safety push as regulatory capture.
“I have not come across a credible, step-by-step documentation of how exactly AI might move from self-recursively improving AI to killing every single human……would likely only wind up serving Anthropic and OpenAI; it’s what regulatory capture looks like in action.”
Merchant’s supposition brings to mind the part about money.
So What About Money?
Sam Altman confirmed OpenAI will not list this year, again citing safety.
BREAKING: Sam Altman says OpenAI will not go public this year due to "safety concerns."
— Financelot (@FinanceLancelot) September 12, 2026
"Given everything happening with safety, right now would be an ill advised moment to go public." https://t.co/kuGrWOU1fu pic.twitter.com/J0Xv7mSJV1
A listing forces a company to publish audited accounts in a filing called an S-1.
“Anthropic and OpenAI delaying their IPO, because their S-1 will reveal that they are bleeding money, and have no path to profitability. Solution? “AI slowdown”, so they cut costs for training new models. It has everything to do with IPO, and nothing to do with safety,” Eli David, AI researcher and co-founder of Deep Instinct, speculated.
OpenAI lost $20.9 billion in 2025 on revenue of $13.1 billion, BeInCrypto reported. Banks have since pushed for investment grade credit ratings, nonetheless.
Anthropic does not expect to break even until 2028, and OpenAI not until 2030. Nvidia has guaranteed $105 billion of OpenAI lease obligations, a backstop that lapses once OpenAI earns a solid credit rating.
BeInCrypto reported in August that Anthropic listing timing details pointed to a late September debut.
However, the theory has a hole. Musk folded his AI arm into SpaceX, which listed in June, so SpaceX AI driven valuation questions are already public. He alone has nothing left to disclose.
Why Russia’s Kirill Dmitriev Said No
Kirill Dmitriev, who runs the Russian Direct Investment Fund (RDIF) and serves as a special representative of President Vladimir Putin, has dismissed the campaign to slow down AI.
“Can’t put genie 🧞♀️ back in bottle,” he said.
In short, it is already too late, with state outlets TASS and Izvestia recasting that as a flat declaration that slowing AI is impossible.
Moscow has little reason to agree. A similar idea appears earlier from President Vladimir Putin (December 2023), who said it is “impossible to stop this development” of AI, including superintelligence.
“If we ban something, it will simply develop elsewhere, and we’ll fall behind.”
That is the same logic Dmitriev is running two and a half years later.
Russia placed 28th of 36 countries in Stanford’s global AI index, far behind the US and China.
Its imports of graphics chips and AI hardware fell 84% in 2024 against pre war levels. Sberbank went hunting for Chinese processors in May to keep its own model running.
Amodei’s plan widens that gap by design. It calls for denying advanced chips to authoritarian states and stretching the democratic lead.
A pact to slow down would freeze Russia in last place. Refusing costs Moscow nothing, because the hardware it needs is blocked either way.
Anthropic has promised outsiders a badge and a desk, and OpenAI has promised to think about it.
“That’s a start, but it’s not enough. When you are racing towards a cliff, you don’t just ease up on the gas pedal. You hit the brakes,” Bernie Sanders articulated.
The US Senator from Vermont calls on Presidents Trump and Xi Jinping to negotiate a treaty to pause AI and ban superintelligence before it is too late.
However, David Sacks says China is very unlikely to join a global agreement.