The Deep Dive Pod
Welcome back to another episode of Deep Dive! This week, the AI world accidentally left its front door wide open. Anthropic inadvertently exposed the entire 512,000-line source code of its flagship Claude Code CLI tool through a simple NPM packaging error. It's like finding someone's digital diary, and what developers discovered inside is absolutely wild. We break down the biggest revelations from the leak, including the discovery of "KAIROS", a secret, unreleased autonomous agent mode that literally "dreams" while you sleep. We also uncovered the bug that burned through 250,000 wasted API callrateNNRESET failure alarming 80% ECOs a day, and the users reported. But the drama doesn't stop there, an Axios supply chain attack hit the NPM registry in the exact same time window, creating a perfect storm for developers. Beyond the Claude chaos: Mistral's massive $830M raise to stockpile Nvidia chips, the ShinyHunters cyberattack on the EU Commission's cloud, and the sudden explosion of physical AI agents, Amazon grabbing robotics startups and Agibot rolling out its 10,000th humanoid robot.
6 episodios
Comentarios
0Sé la primera persona en comentar
¡Regístrate ahora y únete a la comunidad de The Deep Dive Pod!