Anthropic sets Claude Code auto mode as default setting

Anthropic announced it will make its "auto mode" the default setting for its AI programming tool, Claude Code, starting August 14, 2026. The change applies to users with Pro, Max, and Team subscriptions and allows the AI agent to proceed with tasks without requiring step-by-step human approval, unless an action is deemed irreversible, destructive, or outside the user's environment.

The company stated that the update is designed to reduce interruptions and allow Claude to work autonomously for longer periods, potentially for hours on large tasks. This shift comes amid broader industry discussions about the safety of autonomous AI systems. Anthropic emphasized that its internal and third-party testing found auto mode to be safer than average human review. In a controlled study with 1,053 paid professional testers, human reviewers caught only 13.6% of dangerous commands, while the auto mode classifier caught 89%.

Anthropic explained that the system routes all tool calls through a safety classifier. If an action is blocked, Claude either finds a safer alternative or asks for user approval. The system will revert to manual approvals if it encounters three consecutive blocks or twenty...