Claude Code makes Auto Mode the default for Pro, Max, Team

Starting August 14, Anthropic is making Auto Mode the default execution mode in Claude Code for Pro, Max, and Team plans. In Anthropic's tests, the safety classifier used in Auto Mode caught 89% of dangerous commands, compared to 13.6% caught by human reviewers approving actions manually.

Why it matters: The 89% vs 13.6% gap is a striking admission that manual approval prompts don't actually catch dangerous agent actions, since humans tend to click through them without real scrutiny. It signals the safety burden for coding agents is shifting from human-in-the-loop review to automated classifiers, changing what 'oversight' even means for AI-written code.