OpenAI Halts Development of Astra Model for Two Weeks Due to Cybersecurity Risks

This decision shows that OpenAI prioritizes cybersecurity in its development process because AI models increasingly pose unexpected access risks to external systems even during research, due to rising cyber-attack capabilities and insufficient existing security measures.