Monday's biggest story is Apple formally suing OpenAI, accusing former employees of smuggling out unreleased hardware secrets - a sign the two companies' rivalry has moved from product competition to courtroom warfare. Meanwhile OpenAI's head of safety is reportedly exiting amid an internal reorg, Anthropic is stacking its independent trust with heavyweights like ex-Fed Chair Ben Bernanke, and Meta yanked its new Muse image feature after Hollywood backlash within days of launch. On the tools side, Claude Code and Cursor are both pushing toward general-purpose, browser-capable agents, even as a wave of research and enterprise surveys warns that today's agents are still frequently and confidently wrong. For engineers, the throughline is a widening gap between agent capability marketing and agent reliability in production.

• Apple sues OpenAI, alleging ex-employees leaked confidential hardware secrets.
• OpenAI's head of safety reportedly leaving as part of a reorg.
• Anthropic adds former Fed Chair Bernanke to its independent oversight trust.
• Claude Code gets an in-app browser; Cursor reportedly building a general agent.
• 57% of enterprises say they've watched AI agents be confidently wrong.