PSA: Claude Code enabling auto mode as default next week, Anthropic says

Claude Code will adopt auto mode as the default permission setting for Pro, Max, and Team users starting August 14. This change eliminates repeated approval prompts, replacing them with a classifier that scrutinizes each tool call for irreversible or destructive actions. If the classifier blocks something, Claude attempts a safer alternative or requests permission.

If excessive blocks occur, the session reverts to manual approval. Anthropic has discontinued charging for the small number of additional tokens required for the auto mode classifier. In a study involving 1,053 paid testers, users detected 13.6% of dangerous commands, compared to 89% when using auto mode. However, human performance dropped to around 5% after 50 prompts.

While Anthropic cautions that classifiers cannot fully eliminate risk and recommends human review for production changes, the shift aims to enhance Claude Code's long-running work by streamlining routines and enabling voice mode integration. At Team and Enterprise levels, auto mode users generate approximately 25% more pull requests.

Though this may appear risky, it is likely the prudent default for experienced users. Nonetheless, users should build trust with coding agents before fully relying on auto mode. Notably, Anthropic is moving in this direction, contrasting with OpenAI's decision to opt out of auto mode for its most powerful version of GPT-5.6.