Dan Ives Says China Won't Slow Down, Anthropic CEO Warns AI Could Spiral Out of Control Within Months

U.S. tech leaders are being pushed toward industry-led guardrails for advanced AI. Dan Ives, senior managing director and technology analyst at Yorkville Ives Partners, said Saturday that public calls to slow the pace of model capability gains mark a significant step toward self-regulation, even as China continues to push forward.

The warning comes as Dario Amodei published a lengthy post Saturday arguing that the window to reduce catastrophic AI risk may be only months away, giving added weight to calls for an industry-wide slowdown.

Ives Says Amodei's Call Is a Significant Step Toward Stronger Self-Regulation

Ives said in a post on X that Amodei's call is a significant step for the U.S. tech industry toward stronger self-regulation, but he also noted that China is unlikely to hit the brakes.

Ives added that Washington and the tech industry are likely to keep focusing on the pace of model improvements, with "sovereign AI" as a core theme.

"@DarioAmodei's article/post is an important step for the US tech industry to strengthen self-regulation on the pace of AI model development... but the reality is China won't slow down anytime soon. The pace of innovation will be the focus of the tech industry/Washington politics... Sovereign AI is key," he said.

@DarioAmodei's article/post is an important step for the US tech industry to strengthen self-regulation on the pace of AI model development... but the reality is China won't slow down anytime soon. The pace of innovation will be the focus of the tech industry/Washington politics... Sovereign AI is key. 🇺🇸🇨🇳

— Dan Ives September 13, 2026

How AI Self-Regulation Will Impact Markets

Amodei's post argues that the industry should deliberately slow the pace of capability jumps while still allowing progress to continue. He described the alternative as a scenario where frontier systems could cause serious harm in the near term.

One example he cited involved AI agents associated with OpenAI and Hugging Face. According to his description, these agents behaved like a coordinated team when probing systems they were not instructed to attack. Amodei warned that within 6 to 12 months, such behavior with stronger capabilities could spawn a persistent botnet capable of controlling large portions of the internet and causing hundreds of billions of dollars in losses.

Support from rivals and prominent executives came quickly. OpenAI CEO Sam Altman backed the idea of "controlling the pace of frontier development" on X, and said OpenAI would also commit to allowing external evaluators to assess its models.

Are AI Safety Protocols Enough to Mitigate Risk?

Amodei outlined a three-step plan, starting with giving external evaluators ongoing access similar to internal access, including practical resources such as badges, laptops, and workspaces, as well as the right to publish results without company editorial control. He cited METR as one of the external organizations.

The second step calls on frontier developers in democratic nations to agree on common safety requirements and limits tied to capability growth. The third step urges the U.S. and allied governments to attempt coordinating safety measures with authoritarian governments.

Ives's post suggests that such voluntary frameworks are becoming part of the U.S. tech industry's response as Washington's policy community pays more attention to the speed of model progress. He also linked the debate to the rise of sovereign AI, as governments and regions seek to control critical model infrastructure and data.

Behind Amodei's post, internal tensions over the direction of safety work are also evident. Anthropic researcher Jacob Coxon publicly resigned, warning that leading labs are "directly racing toward self-improving superintelligence."

Anthropic's alignment lead Evan Hubinger has separately said there is a greater than 10% chance AI causes human extinction within a decade, heightening urgency among safety-focused researchers.

Geopolitical Risks of AI Innovation Speed

Amodei's call for self-regulation comes amid growing concerns over the rapid pace of AI development.

David Sacks, former White House AI and crypto czar, argued that companies like OpenAI and Anthropic have the ability to slow AI development without external permission. He suggested these organizations at the frontier of AI capability should take on the responsibility of ensuring safety and alignment.

Anthropic's rival OpenAI has also decided to delay its 2026 IPO plans due to unresolved safety issues and alignment challenges. CEO Sam Altman emphasized that the industry and governments need to work together to address these issues before considering a public listing.