What happened
Anthropic co-founder and CEO has argued that the pace of improving AI model capabilities must be slowed, according to reports.
The CEO clarified that this does not mean stopping model training or technological progress.
Instead, the aim is to give companies enough time to align and safeguard their models, with third-party evaluators confirming that this has been done.
Why it matters
The call comes from the head of a leading AI developer, which gives the argument weight in a debate that often pits safety concerns against competitive pressure.
Framing the slowdown as a matter of timing rather than a halt suggests a middle path: keep building, but build in room for alignment work and outside verification.
If third-party evaluation becomes a expected step, it could reshape how AI firms release models and how much scrutiny they face before deployment.
Key facts
Anthropic's co-founder and CEO said the speed of improving AI model capabilities must be slowed.
The CEO said this does not mean stopping model training or technological progress.
The goal is to ensure companies have enough time to align and safeguard their models.
Third-party evaluators should confirm that this has been done.
What to watch next
Whether other AI developers publicly back the idea of slowing capability gains, or continue to prioritize speed.
How the role of third-party evaluators develops, and whether their confirmation becomes a standard part of model releases.
Whether the distinction between slowing capability growth and halting progress holds up as the industry responds.