OpenAI's upcoming Astra model demonstrates advanced cybersecurity capabilities that approach the company's highest risk threshold. Internal safety testing reveals the model can autonomously navigate complex digital systems, prompting OpenAI to temporarily pause parts of its development. This milestone highlights the rapid convergence of frontier models with sophisticated offensive and defensive cybersecurity execution.
Opening Kapyn…