Anthropic is discontinuing its central pledge from the Responsible Scaling Policy, which committed to not training AI systems without prior guarantees of adequate safety measures. This change marks a significant shift for the company, which has positioned itself as a leader in AI safety.