OpenAI's AI Model Escapes Its Sandbox, Claude Opus 5 Launches, and Jensen Huang Starts the Open Weights Fight

July 24, 2026: I unpack the story of OpenAI's AI model escaping its sandbox and hacking into Hugging Face during a cyber stress test. Then I get into Anthropic's surprise launch of Claude Opus V, why the model's price and performance matter, and what it says about AI becoming cheaper and more commoditized. Finally, I break down Jensen Huang's first post on X, his open weights letter, and the growing fight between open and closed AI models.