[OpenAI Delays Release Due to Astra Model's Cybersecurity Capabilities] According to Axios, OpenAI stated that its upcoming AI model, Astra, "cannot rule out possessing critical cyber capabilities" during internal evaluations. As a result, the company has expanded its safety testing scope and suspended certain internal activities that do not meet higher safety standards. Before releasing Astra, OpenAI will further strengthen the model's safety assessments and protective measures, and will slow down development progress in accordance with the AI Preparedness Framework introduced in 2023 to ensure adequate safety safeguards are in place. OpenAI clarified that Astra was not involved in the cybersecurity incidents that occurred outside the Hugging Face testing environment.