Anthropic CEO calls for deliberate slowdown in AI model development

Anthropic chief executive Dario Amodei has called on artificial intelligence companies to deliberately moderate the rate at which they advance model capabilities, setting out a three-step framework aimed at pacing development and creating more time to manage its risks.

In an essay, Amodei argued that the industry must slow the pace at which it improves the capabilities of AI models. Even with such a deliberate slowdown, he said, progress will still appear fast, and the time gained should be used wisely.

The intervention places the head of one of the leading AI developers at the centre of an ongoing debate over how quickly increasingly capable systems should be built and released. Amodei's framework is intended to give companies and regulators more room to address the risks that accompany more powerful models.

He did not specify how the proposed slowdown would be implemented or enforced. The essay sets out a broad direction rather than detailed mechanisms, leaving open how firms might coordinate on pacing their advances.

The call comes as AI companies continue to compete on model performance, with each new generation of systems drawing attention to both their potential benefits and the challenges of overseeing them.