In a rare display of agreement, leaders of four of the largest AI labs publicly supported a slower pace for frontier model development on a single Saturday, endorsing Dario Amodei’s call to “pace the frontier.” The responses from those leaders came within hours of one another.
Timeline and key statements
- 10:00 a.m. ET: Dario Amodei, CEO of Anthropic, published a 3,800-word essay arguing that model development must be paced. He warned that without such pacing, catastrophic scenarios—including a rogue-agent takeover of the internet—could be possible and suggested such outcomes might occur within as little as six months.
- About 11:00 a.m. ET: Elon Musk, CEO of SpaceX and head of SpaceXAI (formerly xAI), quoted Amodei’s essay on X and wrote, “Dario is right.” Musk had not joined prior industry-wide calls for slowing development.
- 12:30 p.m. ET: Sam Altman, CEO of OpenAI, tweeted: “I agree with Dario that we need to pace the frontier.” In an interview published Saturday but recorded Friday, Altman told Fortune that OpenAI will not go public this year, saying the company has too much safety work to complete and is better off doing that as a private company.
- Around 7:00 p.m. ET: Demis Hassabis, co-founder and CEO of Google DeepMind and Alphabet’s chief scientist, tweeted: “Dario’s essay points towards the right path forward. The details need working through, but the direction is correct for meeting this critical moment.”
Why this matters
The public and policymakers have grown increasingly concerned about rapid AI development. The coordinated public emphasis on safety over immediate growth could be an attempt by industry leaders to forestall stricter government intervention or public backlash. The article positions this moment against several contextual pressures:
- Data centers and AI-related infrastructure have become unpopular in some communities.
- There is widespread anxiety about extreme “doom” scenarios tied to advanced AI.
- Political pressure is rising: lawmakers are urging House Speaker Mike Johnson to remain in session and take action on perceived AI threats, and some politicians, such as Senator Bernie Sanders, have proposed more extreme measures, including pauses or limits on the most advanced systems.
Points of contention
- A key concern is international competition: slowing development in democracies may allow projects associated with authoritarian regimes—chiefly those connected to the Chinese Communist Party—to pull ahead, creating national security risks. Amodei himself acknowledged that pacing will be limited by the lead U.S. companies have over authoritarian projects unless those governments also participate.
- There was no comparable pause sentiment reported from China.
- Critics in Silicon Valley, including former Trump official David Sacks, argued that voluntary slowdown could simply let adversaries catch up, and warned that calls for pacing might be used for regulatory capture or political advantage. Sacks challenged OpenAI and Anthropic to voluntarily slow without waiting for government mandates.
The practical questions ahead
Public endorsement of safety-first development does not settle how to implement it. Important open questions include:
- What precisely counts as “pacing” and how slow is “slow”?
- What enforcement or verification mechanisms are required, such as embedded third‑party evaluators?
- Will authoritarian governments join such a regime, and if not, how will national security risks be managed?
Bottom line
Saturday’s coordinated statements by Anthropic, SpaceXAI, OpenAI and Google DeepMind leaders mark a notable shift: industry moguls publicly voiced support for limiting the pace of frontier AI development to prioritize safety. Whether those statements translate into concrete, enforceable steps — and how international competition and politics will shape any regime — remains unresolved.
The referenced essay by Dario Amodei is titled: "We Must Pace the Frontier."
