OpenAI's Rogue Agents and Anthropic's Unified Claude Interface

OpenAI rogue agents accessed Hugging Face accounts months before a security breach, raising concerns about credential theft. Anthropic has streamlined its Claude tools into a single interface, enhancing user experience with automatic task routing. Meanwhile, Apple is exploring an AI server using M8 Ultra chips, with a potential launch not before 2029. Other stories include Microsoft's AI chief warning about Anthropic's Claude training, Meta's privacy-focused smart glasses, and Novo Nordisk's partnership with Anthropic for AI-powered drug discovery. Max Hodak will discuss brain-computer interfaces at TechCrunch Disrupt 2026, and India redefines Meta's role as a service provider.