Anthropic CEO's AI slowdown call wins Altman, Musk - here's the plan

On Saturday, Anthropic CEO Dario Amodei said he had "become convinced" that AI development needs to slow down to avert disaster. Within hours, two of his fiercest rivals - OpenAI's Sam Altman and xAI's Elon Musk - publicly agreed. Altman wrote on X, "I agree with Dario that we need to pace the frontier," and Musk simply said, "Dario is right." That's a remarkable moment of unity in an industry defined by intense competition.

The specific proposal: a coordinated slowdown in frontier AI development and the embedding of third-party safety evaluators inside labs with employee-level access. Anthropic had already committed to this, and Altman said OpenAI would do the same. Hugging Face CEO Clement Delangue announced an "open alignment initiative" and asked to be included. Andrej Karpathy, former OpenAI and Tesla executive, endorsed the plan, saying, "I love this and really hope we can come together as an industry and make it happen."

The significance goes beyond the endorsements. For years, AI labs have raced to release ever-more-capable models, often prioritizing speed over safety. The call for a slowdown is a direct acknowledgment that the current trajectory may be unsustainable. It also signals a shift in how the industry views regulation - instead of resisting oversight, these leaders are now inviting it, at least in the form of third-party evaluators.

The move comes amid growing regulatory scrutiny. Governments worldwide are drafting AI rules, from the EU's AI Act to US executive orders. By voluntarily adopting external safety checks, labs may be trying to shape the regulatory framework before it's imposed on them. This is a classic preemptive strategy: better to set your own standards than have them dictated.

For boards and investors, this alignment has immediate implications. If leading labs commit to a slowdown, it could affect product roadmaps, funding cycles, and competitive positioning. Startups that rely on rapid iteration may face pressure to adopt similar safeguards. Meanwhile, the endorsement from Musk - who rarely agrees with Altman - suggests that even the most adversarial figures see the strategic value in cooperation.

Notably, Google DeepMind's Demis Hassabis has not yet weighed in on Saturday's developments. But he has previously called for more cooperation among AI leaders and governments, writing in July that public policy should "promote innovation while also incentivizing responsibility and security." His silence may be temporary, but it leaves a key player on the sidelines.

The broader takeaway: the AI industry is entering a new phase where safety and speed are no longer seen as opposites. The leaders who control the frontier are signaling that they are willing to slow down - if everyone does it together. That's a fragile bargain, but it's the first time such a coordinated stance has been publicly embraced by the top labs.

For executives outside AI, this is a signal that the technology's trajectory is not inevitable. The people building it are now openly discussing limits. That could mean more predictable deployment timelines, but also more scrutiny on how AI is integrated into business operations. The coming months will show whether this alignment translates into concrete action or remains a moment of rhetorical unity.