Dario Amodei, CEO of Anthropic, published a long essay calling for the AI industry to slow down. Several prominent figures — including Sam Altman, Elon Musk and Demis Hassabis — publicly nodded in agreement. Days later, leaks surfaced in test accounts referring to OpenAI’s GPT‑6 Sol and Anthropic’s Claude Opus 5.2.
What happened
- Dario Amodei released an essay urging developers and companies to decelerate the pace of advanced AI development.
- Sam Altman (OpenAI), Elon Musk and Demis Hassabis (DeepMind) signaled public agreement with the call for caution.
- Within days, test accounts began showing evidence of new models or capabilities labeled GPT‑6 Sol (OpenAI) and Claude Opus 5.2 (Anthropic).
Testers reporting on these appearances noted faster response times, lower token costs and features that enable models to check their own outputs. In addition, Greg Brockman of OpenAI briefly teased a breakthrough in pure mathematics.
Timeline and apparent contradiction
The essay advocating caution was published, and six days later the essay’s author was associated with a release of a next‑generation model. That close timing has prompted questions: the public advocacy of slowing development appears at odds with rapid product advancement from the same actors.
Why it matters
The episode highlights a recurring competitive dynamic in the AI field: organizations publicly support restraint while privately racing to deploy more capable systems. Public calls for caution can function strategically, encouraging competitors to pause first while allowing the signatories to move ahead.
Conclusion
The contrast between the rhetoric of restraint and near‑simultaneous model rollouts suggests that appeals to slow down can serve tactical purposes in a competitive environment. Observers should treat public statements about caution as one input among many when assessing the pace and direction of AI development.
