NEWS · MODELS · #4
OpenAI safety overview: GPT-6 Astra reaches 'Critical' cybersecurity level
According to OpenAI, GPT‑6 Astra is their most capable broadly deployed model and is the first model to reach the Critical level of cybersecurity capability under OpenAI’s Preparedness Framework. The announcement is framed as a safety overview from OpenAI.
KEY POINTS
- According to OpenAI, GPT‑6 Astra is their most capable broadly deployed model and is the first model to reach the Critical level of cybersecurity capability under OpenAI’s Preparedness Framework.
- The announcement is framed as a safety overview from OpenAI.
- If confirmed, a model being rated Critical under OpenAI’s Preparedness Framework signals increased cybersecurity capabilities and could affect deployment, oversight, and risk management for advanced AI systems.
WHY IT MATTERS
If confirmed, a model being rated Critical under OpenAI’s Preparedness Framework signals increased cybersecurity capabilities and could affect deployment, oversight, and risk management for advanced AI systems.