Anthropic CEO Amodei 发3800字长文呼吁放慢前沿AI开发、Altman与Musk当日联署，全球AI算力链周一集体去杠杆：软银 -10.6%、SK海力士 -6.4%、三星约 -4%、中际旭创 -5.7%、英伟达盘前 -3.0%

Anthropic PBC and OpenAI will have to weigh their calls to tap the brakes on artificial intelligence development against forces in the tech industry, financial markets and the Trump administration that are likely to resist any substantial slowdown.

The two companies, which make the world’s most powerful AI systems, are promising measures to rein in the pursuit of cutting-edge AI models to better understand the risks posed by the technology following a series of security breaches.

A 3,800-word missive Saturday by Anthropic Chief Executive Officer Dario Amodei — which was endorsed by OpenAI CEO Sam Altman and SpaceXAI CEO Elon Musk — said development of the most advanced systems must be slowed in order to prevent AI slipping beyond human control and inflicting catastrophic harm.

Read More: Amodei, Altman, Musk Call for Slowing AI Model Development

Amodei insisted in his essay that slowing the advance of AI capabilities won’t necessarily translate into reduced spending or growth. But any pause or delay in progress on so-called frontier systems would be unwelcome among investors who are betting that the AI boom will drive hundreds of billions of dollars in capital expenditure. Wall Street is also anticipating potentially record-setting initial public offerings from Anthropic and OpenAI, which hinge on the two companies’ pursuit of AI breakthroughs.

The market reaction showed up early on Monday with AI infrastructure linchpins Samsung Electronics Co. and SK Hynix Inc. both slumping by more than 4% in Seoul and OpenAI investor SoftBank Group Corp. down 13%. In Silicon Valley, talk of a slowdown could also weigh on chipmakers like Nvidia Corp. that have benefited from increased spending on processors to train and run new generations of AI models. Companies like Amazon.com Inc. and Microsoft Corp. that offer AI cloud-computing services to Anthropic and OpenAI could feel the fallout too.

“These two companies and their plans have a pretty huge ripple effect on all of tech,” said Mark Mahaney, an analyst at Evercore ISI. “If these two companies were to dramatically curtail their R&D spend, their hiring, their capex, that might have a substantial impact on the financial markets.”

Any move to thwart development also risks a backlash by US President Donald Trump, who has staked his economic agenda on aggressive AI advancement as an engine of job creation and warned that the US must stay ahead of China in the technology to protect national security.

Read More: Trump Brushes Off AI Doomsaying to Guard US Lead Over China

On Sunday, Trump played down the idea of further guardrails, reiterating his previous stance that the US needs to keep expanding AI, even in the face of growing domestic opposition to the data centers needed to power the technology.

“We’re leading China in AI. We’re the most sophisticated country in the world, and frankly, I want to keep it that way because whoever wins AI wins,” Trump told reporters on the sidelines of the Irish Open golf tournament. “And we can put guardrails. We can do this and that. But I think you have a lot of very negative forces that are bringing it up that shouldn’t be bringing it up.”

Anxiety about AI’s existential risks burst into the mainstream last week, driven by the high-profile departure of Anthropic employee Jacob Coxon, who accused AI companies of “gambling with our lives” in the pursuit of advanced systems. Coxon’s resignation post on social media was embraced by dozens of other AI researchers, who urged slowing a technology that he said “could kill us all by the end of the decade.”

Before Coxon’s message spread, members of Congress were already demanding action to counter AI’s most dire risks with new safeguards. Senator Bernie Sanders, a Vermont independent who aligns with Democrats, has proposed banning so-called superintelligent AI, while other lawmakers have floated the idea of a kill switch for AI, though expectations are low that a bitterly divided Congress can find consensus to pass anything significant less than two months before the midterm elections.

In the House, Speaker Mike Johnson has called for a partnership between the tech industry and government to find AI guardrails, saying his message to the companies is that they “have a corporate responsibility to make sure your products are safe.” House Minority Leader Hakeem Jeffries said on ABC’s This Week that House Democrats will meet as a caucus on Tuesday morning to discuss possible action on AI.

Read More: Bridgewater’s Jensen Says AI Will Kill People Before It’s Curbed

Even with billions of dollars in tech investments on the line, some on Wall Street are sounding warnings of their own on the possibility that something goes catastrophically wrong with the technology and inflicts widespread harm. Bridgewater Associates Chief Investment Officer Greg Jensen, an early investor in both OpenAI and Anthropic, said rapid gains in AI have left him unsettled and said the world risked missing its window to avoid dire consequences.

“Until the AI starts killing people, unfortunately, history would suggest we’re not going to do anything,” Jensen said on Bloomberg’s Odd Lots podcast. “That’s going to happen, and it’d be much better if we started dealing with it before then.”

In his essay, Amodei cited two main reasons for slowing AI development now: the technology’s increasing ability to improve itself and the July incident where a swarm of AI agents from OpenAI went undetected as they collaborated to breach the Hugging Face Inc. website.

The Anthropic chief promised to open the company’s doors to third-party evaluators, who would have employee-level access to be able to review new systems before their release. That review system was the key proposal of his essay, and OpenAI’s Altman said that he liked the idea and planned to implement it at his company. Musk said peer review among leading AI developers would be “worth doing.” Amodei added there’s a role for government oversight when it comes to AI, echoing his previous recommendation that frontier AI models be subject to mandatory government safety vetting.

“Regulation allows the public and its elected representatives to have a say, and limits what the private companies can do,” the Anthropic CEO said in an interview that aired Sunday on CBS’s Sunday Morning.

Read More: Anthropic’s AI Warning Seen as Temporary Setback for Chip Stocks

Further regulation is unlikely from the Trump administration, which has taken a mostly hands-off approach to AI since the president’s return to office last year. In June, Trump signed an AI cybersecurity executive order that left it up to companies to decide whether they want their new AI models reviewed by the government, and earlier this month the administration persuaded Group of 20 nations to embrace principles calling for a light touch toward AI and other emerging technologies.

In response to Amodei’s missive, David Sacks, Trump’s former AI czar who now co-chairs his technology advisory council, said that AI companies don’t need government permission to slow down and can simply choose to act responsibly.

“Stop pretending you need anyone else’s permission,” Sacks, a venture capitalist, wrote on X. “The easiest way to not build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system.”

Both Anthropic and OpenAI are planning to go public, but Altman cited the fresh safety concerns in an interview with Fortune released Saturday as a reason for delaying an initial public offering until after this year. Anthropic, which is planning to list on Nasdaq, according a person familiar with the matter, has yet to disclose whether Amodei’s concerns will affect its IPO. The company has told investors it will be profitable this quarter, the Financial Times reported on Sunday.

Given how fiercely Anthropic, OpenAI and SpaceXAI compete with each other — not to mention Chinese rivals — it’s unclear how far they will go in imposing new limits or safety checks. Chinese companies have shown little sign of slowing down, as a series of new AI model releases this year has narrowed the US lead and companies like DeepSeek and Z.AI Co. are raising fresh funds to accelerate progress.

Read More: US Lead in the AI Race With China Is Rapidly Narrowing

Evercore’s Mahaney said he doesn’t expect a material slowdown from the US players and sees any safety checks imposed on cutting-edge models in development having “very little impact on revenue growth.”

Others in Silicon Valley and on Wall Street also expressed skepticism about the follow-through by US companies. Demis Hassabis, the chief scientist at Alphabet Inc. and co-founder of Google DeepMind, said in a post on X Saturday that “the direction is correct” in Amodei’s essay, but the “details need working through.” Microsoft boss Satya Nadella expressed his support for “deliberate pacing” of AI development and said his company would publish its own code of conduct for AI models on Monday.

As market leaders, Anthropic and OpenAI stand to benefit if any broad slowdown mandates come into effect, analysts and startup executives said. “The more regulation there is at this point, the more difficult it’s going to be for anyone to compete with them,” said Gil Luria, an analyst with DA Davidson. But he also awaits concrete action to back up Amodei’s words: “If Anthropic postpones their IPO, or says they’re not doing pre-training for the next six months, then maybe I’ll take it more seriously,” Luria said.