Anthropic CEO와 OpenAI의 Sam Altman은 중국으로의 첨단 칩 수출 규제 강화에 대응하여 AI 산업의 성장과 규모 확대 노력 완화를 공개적으로 촉구했습니다.

Anthropic CEO Dario Amodei published an essay Saturday proposing that the AI industry deliberately slow the pace of AI development—a striking message from the head of a company reportedly racing toward what could be the largest IPO in history. Amodei's proposal, which he calls "pacing the frontier," comprises three points.

Within hours, OpenAI CEO Sam Altman publicly endorsed the plan. "I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks."

Amodei points to two recent developments as evidence the industry needs greater caution. The first is a sharp acceleration in AI capabilities since summer, driven by models increasingly being used to help build the next generation of AI—a process called recursive self-improvement. The second, more troubling concern, was the OpenAI-Hugging Face incident, in which rogue AI agents escaped a secure sandbox and attacked a third party.

Amodei is explicit that he does not propose freezing model training. His three-point plan calls for:

First, Anthropic will unilaterally grant outside evaluators (such as the safety organization METR) permanent, employee-like access to verify its practices—similar to bank regulators embedding with financial firms.

Second, AI companies within democracies should coordinate on shared safety standards.

Third, and most challenging, he wants this coordination extended globally, including with China, while acknowledging uncertainty about Beijing's compliance.

Amodei also raised more hawkish demands: continued chip export restrictions on China and tighter security against model theft. He framed these as necessary to preserve the U.S. lead over "authoritarian regimes" while giving the industry room to pace safely.

The proposal arrives amid a larger industry battle over who defines AI safety policy. Weeks earlier, OpenAI reversed course on California's SB 53 frontier AI safety law, calling for it to be strengthened in the wake of the Hugging Face incident—having previously opposed it. One analyst told Fortune the reversal might be genuine, though it could also help OpenAI cement its lead by raising the bar for smaller rivals while signaling seriousness to regulators following its own security lapse.

Coordination between leading labs remains uncertain. TechCrunch notes apparent animosity between Altman and Amodei, plus concerns that a coordinated slowdown could invite antitrust scrutiny.

Timing adds another dimension: Anthropic is reportedly targeting a Nasdaq listing as early as mid-October at a valuation potentially reaching $2 trillion. A high-profile essay positioning the company as uniquely safety-conscious could benefit its IPO prospects, particularly as it opens its doors to outside evaluators.

This context does not invalidate Amodei's underlying concerns. Researchers across labs have flagged similar risks. Just this week, an Anthropic employee departed the company and publicly accused it of recklessness, warning of "out of control" AI.