Anthropic has announced that it will make the automatic execution mode (auto mode) the default setting for Pro, Max, and Team plan users in its coding assistance tool, Claude Code. Consequently, users on these plans will automatically operate in auto mode starting from new sessions. For users who have already manually changed their default settings, a prompt may appear to confirm the switch.

Auto mode is designed to reduce confirmation prompts for the user while utilizing a classifier to detect and block irreversible or destructive operations. According to a study conducted by Anthropic involving 1,053 testers, humans detected only 13.6% of dangerous commands, whereas auto mode blocked 89%, demonstrating improved safety. Anthropic stated that automatic execution enables long-term autonomous task performance without interrupting the user's workflow.

Meanwhile, in enterprise environments such as Claude Enterprise, Claude API, Amazon Bedrock, and Google Cloud, an opt-in format will be maintained for the time being to allow administrators time for review. Anthropic plans to make this the default setting for deployments via these cloud partners within the next month.

Source: