Anthropic announces Claude Code auto mode becomes default August 14

On August 8, Anthropic announced that Claude Code will switch to auto mode as the default permission setting for Pro, Max, and Team plans starting August 14, unless a user or admin has pinned another mode. Auto mode uses a classifier to approve routine actions and flag irreversible or out-of-bounds ones.

Anthropic's study of 1,053 testers found humans caught 13.6% of dangerous commands. Auto mode aims to reduce operational risk through classifier-based gating while preserving Claude Code's autonomous agent flexibility. The mode is designed for multi-step long-session workflows where the classifier judges before each action whether human approval is needed.

The auto mode change does not affect core model capability. It only changes Claude Code's client-side permission management. Anthropic said users and teams can switch back to other modes, such as fully manual or pre-approved fixed permission lists, through system settings.

The default switch is a key milestone for Claude Code in enterprise deployments. In enterprise scenarios, automatic approval of routine actions can significantly improve developer productivity, while classifier-based flagging of irreversible actions reduces high-risk events such as accidental file deletion or wrong code commits.