
Published on September 2, 2026 at 10:31 Comment 2 𝕏 Print © OpenAI / 01net OpenAI revealed that Astra, its future model intended for ChatGPT, crossed the “critical” threshold for the first time during a safety evaluation. The AI is capable of discovering unknown vulnerabilities and independently mounting cyberattacks against protected systems, without human intervention. After taking a series of precautions, OpenAI intends to launch the AI in the near futu





