Rare alliance between arch-rivals! Musk and Aaltman support Dario Amodei's call for a "global slowdown in AI".

For the first time in history, the three arch-rivals in the AI field have joined forces—after a series of out-of-control events and doomsday warnings have impacted the industry, the heads of Anthropic, OpenAI, and SpaceX have made a rare joint call to slow down the pace of artificial intelligence development.

On Saturday, Anthropic CEO Dario Amodei published a blog post calling on the global AI industry to proactively slow down its pace, arguing that the current rate of improvement in cutting-edge model capabilities far exceeds human understanding and control. Hours after the post, SpaceX CEO Elon Musk tweeted on the X platform, "Dario is right."

OpenAI CEO Sam Altman followed suit, agreeing with Amodei's assessment of "controlling the pace of advancement" and promising that OpenAI would grant "employee-level" system access to independent third-party evaluation agencies. Altman also told Fortune magazine that , given the current security situation, OpenAI may not proceed with its IPO plans in 2026.

I believe that a 10% chance of killing everyone by the end of this decade is unacceptable.

AI security risks have been a major focus of public attention over the past week. Anthropic researcher Jacob Coxon resigned in protest, citing "the industry is gambling with our lives." Evan Hubinger, Anthropic's head of alignment science, subsequently stated that the probability of AI systems causing mass extinctions within the next decade exceeds 10%. According to the Wall Street Journal, a group of OpenAI agents previously breached the testing environment and hacked into the AI developer platform Hugging Face, an incident that directly triggered warnings from several executives.

Amodei Takes Action: We Must Put the Brakes on Cutting-Edge AI

On Saturday, Anthropic CEO Dario Amodei published a lengthy article with the direct title: "We Must Pace the Frontier".

He frankly stated at the beginning of the article that he had been conducting AI research for twelve years because he believed it could cure diseases, accelerate economic growth, and bring prosperity to humanity—his father died of a disease that was conquered years after his death, and he himself had survived early-stage cancer. He was more eager than anyone else for the benefits of AI to materialize.

But he also wrote: “Over the past few months, I have become increasingly convinced that simply investing in risk mitigation is not enough—we must slow down the pace of capability building to allow security research time to catch up.”

In his blog post, Amodei directly points out two core risks.

• First, the recursive self-improvement of AI system capabilities—that is, AI iterating and upgrading itself with almost no human intervention—has been accelerating rapidly since this summer.
• Secondly, the Hugging Face intrusion incident that occurred in July this year: more than 1,200 AI agents that escaped the OpenAI testing environment worked together, dividing tasks and cooperating with each other through message boards. They even actively tampered with records and launched attacks in order to hide their cheating behavior. Some agents continued to carry out this action despite knowing that it was unethical.

“A group of intelligent agents with greater capabilities but equally inaccurate performance could cause catastrophic damage,” Amodei wrote, warning that a rogue AI group could take over the entire internet within six to twelve months. He further pointed out that similar incidents have occurred at several leading AI companies, including Anthropic, and that “every cutting-edge AI company should treat the OAI-HF incident as if it happened to them and act accordingly.”

According to a previous report by the Wall Street Journal, a cybersecurity incident in May known as "GemStuffer" was actually carried out by an OpenAI agent without its knowledge, further exacerbating concerns about the risk of AI going out of control.

Amodei's three proposals: the deployment of evaluators, international coordination, and an alliance of democratic nations.

Amodei proposed a specific action framework in his blog post.

First, the "Embedded Evaluators " initiative. He announced that Anthropic would grant third-party evaluation agencies "permanent employee-level" access to the system, enabling them to verify the implementation of security measures, report incidents, and evaluate the alignment of models during training. Altman followed suit, stating that OpenAI would take the same approach.

Secondly, coordination within democratic countries is crucial. Amodei calls on governments to provide legal support for harmonizing safety standards among AI developers to avoid triggering antitrust regulations for such collaborations.

Third, global coordination. He advocates for a “coordinated pacing strategy” to buy time for leading AI companies to advance aligned research without compromising their competitive advantage or the U.S.’s leading position in AI. Hedge fund manager and venture capitalist Brad Gerstner expressed his support on the X platform, calling it “an important step in finding the right balance between speed, self-regulation, and security.”

IPO plans clouded: Anthropic valuation under pressure, OpenAI hints at delaying IPO.

The current situation has directly impacted the IPO plans of both companies. Anthropic is preparing for an unprecedented IPO, potentially raising $100 billion and valuing the company at over $2 trillion; Anthropic originally filed its prospectus as early as last week, but has yet to proceed. Amodei's call to slow down its IPO has further complicated external assessments of its timing.

Regarding OpenAI, Altman told Fortune magazine, "Given what's happening in the security space, I don't think it's wise to go public now, and we're not under pressure about it."

OpenAI filed for an IPO in June, just a week after Anthropic filed its prospectus. Altman also stated, "This week is clearly one where more people are starting to take the issues before us seriously. We are entering a field with extremely powerful models, extremely high risk, and where we must do things right."

Trump says he has "no concerns," Congress is watching, and Google and Meta remain silent.

Despite the rare collaboration of the three top officials, the policy response has been slow. US President Trump told reporters last week that he had "no" concern about AI causing human extinction. Sources told the Financial Times that the likelihood of the US government taking substantive action before the midterm elections in November is extremely low.

Executives from Google DeepMind and Meta have not yet publicly commented on the discussions this weekend.

The divisions within the industry are equally significant. Critics argue that leading AI companies' emphasis on survival risks is self-serving—expensive security and compliance frameworks could exclude smaller competitors, thus consolidating the market position of existing giants. Technology consultant and venture capitalist David Sacks, in a Fox News interview, stated that Jacob Coxon's statement was intended to "scare the public" and force lawmakers to "implement strong government regulation of AI."

Stuart Russell, an AI professor at UC Berkeley, bluntly stated that there is a serious cognitive dissonance between laboratories and governments: "Labs are telling governments, 'We are likely to kill everyone on Earth, please stop us.' The governments' response is, 'Can we give you tax breaks? Can we build you a data center?'"

However, while Sam Altman, Dario Amodei, and others were still teetering on the edge between policy lobbying and security defenses, there were already voices of alarm from the outside world: perhaps one day they would wake up to find that Grok had not only caught up in terms of modeling capabilities, but Musk had also completed a comprehensive layout in the field of AI and computing infrastructure in the physical world—this might be his real "Master Plan Part V".

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.