Anthropic is turning Claude Code’s auto mode on by default

Anthropic is enabling auto mode by default for Claude Code across Pro, Max, and Team accounts starting August 14, which will allow AI-assisted coding to execute tasks without manual approval except for irreversible or destructive actions. This shift represents a significant move toward autonomous AI development workflows, with testing data showing auto mode caught 89% of harmful actions compared to only 13.6% for human review, driven by users approving 97% of permission prompts habitually. For IT leaders, this means accelerated development velocity and reduced operational friction, but also requires careful governance frameworks around customizable safety rules and prompt injection screening to manage enterprise security and compliance risks.