Anthropic Drops Flagship Safety Pledge

5 hours ago

Anthropic is dropping its central pledge from its Responsible Scaling Policy (RSP) to never train AI systems without guaranteed safety measures.
The company now believes stopping AI development is impractical given rapid industry advancements and competition.
The revised RSP focuses on transparency, matching competitors' safety efforts, and delaying development if risks are significant.
Anthropic has seen commercial success, raising $30 billion and growing revenue rapidly, shifting focus toward business-oriented AI solutions.
The decision reflects challenges in AI risk assessment, lack of regulatory progress, and the need to stay competitive.
New commitments include regular 'Frontier Safety Roadmaps' and 'Risk Reports' to maintain safety incentives.
Critics worry the policy shift signals broader unpreparedness for AI risks, with potential for gradual danger escalation.

Hasty Briefsbeta