OpenAI suspended Astra model development to assess its network security risks

OpenAI announced a two-week suspension of partial training and development of its unpublished Astra model. Early assessments indicate that the model may reach the “critical” network security risk threshold under the corporate readiness framework。