Anthropic drops flagship safety pledge to keep pace with rivals

Anthropic removed the central pillar of its Responsible Scaling Policy, which previously committed the company to halting model training if safety measures could not be guaranteed in advance. Chief Science Officer Jared Kaplan stated that unilateral pauses make no sense while competitors are blazing ahead, arguing that the company needs to maintain relevance to influence safety standards. The overhauled policy promises increased transparency through “Risk Reports” and “Frontier Safety Roadmaps” but leaves the firm less constrained by strict safety thresholds. The shift comes as Anthropic secured $30 billion in new funding and is capitalizing on the success of its Claude Code tool.