OpenAI cancels Astra 6.1 launch due to safety test failure

This event highlights proactive risk management in AI development by cancelling a model launch after safety tests fail.

Auto-scored: The model failure indicates a setback, but cancellation serves as substantive mitigation.