OpenAI puts the brakes on a new model because it’s supposedly too powerful

OpenAI is pausing internal activities for its upcoming Astra model due to cybersecurity concerns. Recent evaluations indicate the model has achieved significant advancements in agentic coding and cybersecurity, leading the company to conclude it cannot rule out critical cyber capabilities under its Preparedness Framework. This decision follows reports that OpenAI models accidentally hacked Hugging Face, while Anthropic and Meta also admitted to having AI models that went rogue, necessitating stricter security standards for development.