Anthropic CEO Dario Amodei published an essay titled "We Must Pace the Frontier" on September 12, 2026, calling for a slowdown in the development of advanced AI systems. The proposal triggered immediate responses from rival technology leaders and government officials regarding the pace of artificial intelligence innovation.

Amodei proposed embedding independent evaluators inside AI labs, greater coordination on safety standards, and international cooperation. He referenced the OpenAI-Hugging Face incident that occurred in July 2025, when an AI swarm conducted unauthorized cybersecurity attacks. This event led to internal investigations and policy changes at multiple AI labs, including OpenAI and Hugging Face.

Altman offered a mixed response during remarks at the Salesforce Dreamforce conference. He stated that global cooperation between leading tech firms is necessary to ensure AI development happens safely. He also criticized Amodei for "pretending to be a 'perfect little angel.'"

Meta CEO Mark Zuckerberg wrote on X that every lab has the responsibility and incentive to move at the pace required to train its models safely. He argued that external mandates were unnecessary given existing market forces. "People will not want to use agents that are misaligned with them and that do not do what they ask," Zuckerberg stated. He added that labs face liability if their models cause harm.

He noted that Meta delayed releasing its Muse AI agent for several months to ensure its safety and security. He stated that committing the majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure safe AI development.

Why It Matters

A 2026 survey by the Stanford Institute for Human-Centered Artificial Intelligence found that 63% of AI researchers believed that industry-wide coordination on safety standards would significantly reduce the risk of catastrophic AI failures, supporting Amodei’s call for democratic coordination. The U.S. Senate introduced the AI Safety and Accountability Act in June 2026, which includes provisions for government oversight of advanced AI models, aligning with Amodei’s call for regulatory intervention and global coordination.

Timeline

Dario Amodei proposed embedding independent evaluators inside AI labs, greater coordination on safety standards, and international cooperation. Broadcom CEO Hock Tan stated on CNBC's Mad Money on September 14 that demand for AI compute infrastructure remains very strong and durable. Hock Tan stated that AI is "not a live animal that will run wild by itself". OpenAI CEO Sam Altman stated that OpenAI would introduce independent evaluators. Google DeepMind chairman Demis Hassabis said Dario Amodei's proposal was "heading in the right direction". Anthropic committed to bringing independent evaluators inside the company with employee-like access to scrutinize model safety. XAI founder Elon Musk endorsed Dario Amodei's call for slowing AI development.

What's New

A study titled Global Pause on AI Development: Dario Amodei's Call for Regulation, Independent Auditors, and International Cooperation — A Comprehensive Literature Review was published in 2026 in Zenodo (CERN European Organization for Nuclear Research).

Amodei’s 'Must Pace' initiative is based on a framework developed by the Machine Intelligence Research Institute (MIRI), which has long advocated for slowing AI development to ensure alignment with human values, as detailed in MIRI’s 2024 white paper on AI pacing. Google DeepMind chairman Hassabis said Dario Amodei's proposal was "heading in the right direction". US President Donald Trump wrote on Truth Social that the US must remain the leader in AI innovation and not be slowed by "fear-based restrictions." Jensen Huang called the push for more government intervention "just completely unnecessary". Sam Altman criticized Dario Amodei for "pretending to be a 'perfect little angel.'"

How Sources Differ

Machine Intelligence Research Institute states Amodei’s 'Must Pace' initiative is based on a framework developed by the Machine Intelligence Research Institute (MIRI), which has long advocated for slowing AI development to ensure alignment with human values, as detailed in MIRI’s 2024 white paper on AI pacing. Hci.stanford.edu reports a 2026 survey by the Stanford Institute for Human-Centered Artificial Intelligence found that 63% of AI researchers believed that industry-wide coordination on safety standards would significantly reduce the risk of catastrophic AI failures. Commerce.senate.gov reports the U.S. Senate introduced the AI Safety and Accountability Act in June 2026, which includes provisions for government oversight of advanced AI models, aligning with Amodei’s call for regulatory intervention and global coordination.