In February 2026 version 3.0 of Anthropic’s Responsible Scaling Policy dropped its commitment to pause developing more powerful models if adequate safeguards were not ready; Anthropic argued that a pause by one company while others continued could leave the world less safe.

Evidence1 sourceCited by 1 pointCreated Nov 5, 2026

Source

PC Gamer: Anthropic ditches its defining safety promise to pause dangerous AI development
Published by pcgamer.comRecorded Nov 5, 2026
Open the original

Supports: Anthropic dropped its promise to pause training when safeguards weren’t ready, the commitment that most set it apart.
- A reply in “Anthropic”