Auto mode is now the default in Claude Code

Starting August 14, new Claude Code sessions on Pro, Max, and Team plans will run in auto mode by default. Users who already set a different default may receive a one-time prompt asking if they want to switch, but pinned defaults remain unchanged. Anthropic is also eliminating the extra token cost for the auto mode classifier on those plans effective today.

Auto mode routes each tool call through a classifier designed to block irreversible, destructive, or out-of-environment actions. When a call is blocked, Claude typically finds a safer way forward or asks the user for approval; if it stalls—three blocks in a row or twenty across a session—it falls back to manual approvals. Enterprise, the Claude API, AWS Bedrock, Google Cloud's Agent Platform, and Microsoft Foundry keep auto mode opt-in for now, with plans to make it default within a month; Enterprise admins can already set it via managed settings.

Testing included internal red-teaming, third-party red-teaming, prompt-injection evaluations, a controlled study with 1,053 paid testers, and analysis of real production sessions. On every metric, auto mode matched or outperformed manual review. Data shows users approve 97% of permission prompts, suggesting habitual clicking rather than careful review. Among Teams & Enterprise adopters, auto mode users ship about 25% more PRs, and companies like Adobe, Nuro, Gusto, and Garner Health already use it as their production default.

The shift enables Claude to work autonomously for longer stretches, making long-running models like Opus 5 more practical for hours-long tasks. Reducing prompt burden increases output and lets developers focus on higher-level decisions.