OpenAI cancels Astra 6.1 launch after model fails safety tests

This event matters as it shows proactive safety enforcement in AI development, preventing potential risks by halting a model launch based on test failures.

Auto-scored: The safety test failure is a minor risk setback, but the cancellation is a substantive mitigation step in governance.