OpenAI plans to release a new artificial intelligence model called Astra soon, but will not make its most advanced cybersecurity-related capabilities broadly available at launch, Bloomberg reports, citing a Tuesday OpenAI blog post. According to Bloomberg, OpenAI did not give a specific release date for Astra. The company said the model is coming “soon,” while setting a staged access plan for its higher-end cybersecurity functions. The first phase will restrict Astra’s ability to perform advanced cybersecurity-related tasks to a group of testers, Bloomberg reports. After that, OpenAI plans to offer access to a larger pool of users for defensive cybersecurity purposes through its Daybreak Blue program. Bloomberg describes Daybreak Blue as a program that lets approved testers use OpenAI’s most capable models with safeguards designed for cybersecurity work. The reported structure keeps the most sensitive cyber capabilities behind an approval process rather than putting them immediately into general release. The key detail is not that Astra has launched; the provided report says OpenAI plans to roll it out soon. Nor does the report give technical benchmarks or a full capabilities list. What it does establish is OpenAI’s intended release posture: advanced cyber features will be gated first to testers, then to a broader approved group focused on defensive use. Who benefits: Approved cybersecurity testers and participants in Daybreak Blue stand to get the earliest access to Astra’s advanced cyber functions. Defensive security teams may benefit later if OpenAI expands access as described. Who's exposed: Users outside the approved tester pool will not initially have access to the most advanced cybersecurity capabilities described by Bloomberg. The report does not provide enough detail to assess broader security risks or commercial availability.