Auto mode is now the default in Claude Code for Pro, Max, and Team plans
02:00 · August 7, 2026 · Claude Blog
Summary
Anthropic is making auto mode the default setting for new sessions in Claude Code on Pro, Max, and Team plans beginning August 14. Users who have already chosen a different default may receive a one-time prompt to switch, while those with a pinned default see no change. The classifier that powers auto mode previously added a small token overhead per tool call; Anthropic has removed charges for that overhead on the affected plans.
Auto mode routes each tool call through a targeted classifier that blocks actions judged irreversible, destructive, or directed outside the user’s environment. When the classifier intervenes, Claude typically identifies a safer alternative or requests explicit approval. After three consecutive blocks or twenty blocks in a single session, the system reverts to manual review. Permission rules created by the user continue to apply, except for broad allow-rules that would bypass the classifier entirely.
The change rests on internal and third-party red-teaming, prompt-injection testing, a controlled study with 1,053 participants, and analysis of production sessions. In the study, participants approved clearly dangerous commands 86.4 percent of the time, while auto mode blocked 89 percent of the same commands. Real-world sessions that passed through manual approval showed more than twice the rate of serious unintended harm compared with auto mode sessions. Prompt-injection evaluations found zero successful attacks against models running auto mode, versus measurable success rates against unguarded or bypass modes.
Teams using auto mode ship roughly 25 percent more pull requests than those relying on manual approvals. The mode is already the production default at several organizations, including Adobe, Nuro, Gusto, and Garner Health, and it serves as the internal default for all Claude Code usage at Anthropic. Enterprise, API, and platform users on AWS, Google Cloud, and Microsoft Foundry can enable auto mode through managed settings; it remains opt-in for these offerings while Anthropic works with partners to extend the default rollout.
Why it matters
Provides actionable implementation details, safety data, and configuration steps for an AI coding tool update directly usable by product teams and builders. Addresses workflow automation, risk mitigation, and observability in long-running AI tasks with specific model references.
