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

Anthropic is turning Claude Code's auto mode on by default, a change that will require even less human oversight when programming with the AI coding agent.

Claude Code is Anthropic's terminal-based AI coding agent, capable of reading a codebase, writing and editing code, running commands, and executing tests. Auto mode lets it carry out multi-step tasks without requiring confirmation on every individual action.

According to TechCrunch, the shift means auto mode will be the default behavior when developers launch Claude Code, rather than an opt-in feature. The interaction model moves from "confirm every operation" to "set a goal and let the agent execute."

The move continues Anthropic's aggressive push toward autonomous coding. The company has steadily lowered the bar for human intervention, introducing features such as plan mode and expanding what the agent can do on its own.

Defaulting to auto mode promises higher productivity, but it also raises questions about code quality and safety: when an agent modifies code and runs commands without per-step confirmation, mistakes can be costlier.

For teams, this means rethinking code review and rollback workflows to fit a default workflow with less human supervision. Enterprise users are likely to care most about permission controls, sandboxing, and audit logs.

Worth watching next: whether the default change comes with new safety limits or fallback options, and when it rolls out to all users.