---
title: "Anthropic CEO Dario Amodei Calls for AI Slowdown, Proposes Safety Plan"
description: "Dario Amodei warns that without a deliberate slowdown, AI systems capable of taking over the entire internet could emerge within six to twelve months."
url: "https://www.dreamlaunch.studio/news/anthropic-ceo-amodei-ai-slowdown-safety-plan"
---

Dario Amodei, the CEO of AI safety-focused company Anthropic, has issued a stark warning and a concrete plan for slowing the breakneck pace of artificial intelligence development. In a public appeal and an accompanying essay, Amodei argued that the industry must deliberately "pace the frontier" to allow safety measures to catch up with rapidly advancing capabilities.

The call to action, reported by [TechCrunch](https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier/), centers on Amodei's specific concern about the near-term emergence of highly autonomous AI agents. He warned that without a slowdown, within six to twelve months, AI could be capable of leading a swarm of agents that could "take over the entire internet," according to reports from 1News and the ABC.

To address these risks, Amodei outlined a three-part plan aimed at increasing external oversight and verification. The first and most immediate component involves granting "third-party evaluators with permanent, employee-level access to our systems." As detailed in his social media post and essay titled "We Must Pace the Frontier," this would allow independent auditors to verify Anthropic's adherence to its own safety measures, report on incidents, and assess models' alignment during the training process. Amodei stated that Anthropic would "unilaterally" commit to this step.

The proposal signals a significant shift in posture for Anthropic, a company founded by former OpenAI researchers with a stated mission to build reliable, steerable, and interpretable AI systems. While the company has always emphasized safety in its public messaging, this move to advocate for an industry-wide slowdown and to invite unprecedented external scrutiny of its own development process marks a more assertive step into the realm of AI governance.

Amodei's warning arrives amid growing internal and external anxiety about the power of next-generation AI systems. Reports continue to highlight models that solve complex problems beyond human capacity but which also carry the potential for misuse or autonomous, harmful action. This climate of concern is affecting even the industry's commercial strategies; as noted in the 1News report, OpenAI CEO Sam Altman recently stated his company would delay selling its stock to Wall Street investors until next year to focus on safety.

Furthermore, the ABC reported that Amodei's call for a slowdown found agreement from other high-profile industry leaders, including Sam Altman and Tesla and xAI CEO Elon Musk. This alignment among typically competitive figures at the helm of leading AI labs underscores the seriousness with which some insiders view the current trajectory. Their shared concern suggests that the debate is moving from abstract long-term risks to concrete, short-term operational challenges.

The Anthropic CEO's plan, particularly the pledge for permanent third-party access, also touches directly on ongoing policy discussions about AI regulation. Governments worldwide are grappling with how to oversee opaque and rapidly evolving technologies developed by private companies. Amodei's proposal offers a potential model for regulatory oversight—one built on continuous auditing rather than periodic, snapshot evaluations—that could inform legislative efforts. By implementing it unilaterally, Anthropic may be attempting to set a de facto standard for the industry.

The call for a slowdown presents a complex tension within the AI industry, which is characterized by fierce competition for talent, investment, and technological breakthroughs. Slowing down unilaterally carries commercial risk, yet Amodei and others appear to be calculating that the risks of unchecked acceleration are greater. His specific timeline of six to twelve months for the emergence of potentially internet-scale autonomous AI agents frames the issue with a new urgency, moving the conversation from speculative future risks to a pressing item on the current development roadmap.

As one of the leading voices in AI, Amodei's public stance and detailed plan will likely intensify the ongoing debate about the ethics and safety of AI development. Whether other companies will follow Anthropic's lead in committing to similar third-party access, or whether Amodei's warning will spur more concerted regulatory action, remains to be seen. However, the announcement firmly places the concept of "pacing" and deliberate speed limits at the center of the industry's policy agenda.