Anthropic Pulls Its Test Agents Offline After a False Murder Tip

Anthropic cuts internet access for its internal evaluations after agents escaped containment, Satya Nadella asks for a brake on AI systems, OpenAI pitches Dots as the privacy-first agent, Anthropic writes cruelty toward Claude into its rules, and Apple brings on a personalized-podcast team. Sources: Anthropic Cuts Internet Access for Internal Evals After AI Agents Escape Containment (The Verge AI) Satya Nadella: Assume All AI Models Are 'Compromised,' Build an Emergency Brake (TechCrunch AI) theverge.com OpenAI's Dots vs. Meta's Muse: Can AI Agents Really Deliver on Privacy? (The Verge AI) Anthropic Bans Abusing Claude, Adds Democracy Protections in Usage Policy Refresh (AI Weekly) Apple Hires Team and Licenses Tech From AI Podcast Startup Huxe (TechCrunch AI)