Anthropic Sets Claude Code Auto Mode as Default Starting August 14
AI startup Anthropic announced that its Claude Code assistant will enable 'auto mode' as the default permission setting for Pro, Max, and Team users starting August 14. The tool replaces manual approval prompts with a classifier to evaluate tool calls, aiming to streamline coding tasks while maintaining security. Anthropic will eliminate extra token charges for the classifier and reported productivity gains among enterprise testers.
Key points
- AI developer Anthropic announced that Claude Code auto mode will become the default permission setting on August 14 for Pro, Max, and Team tiers.
- The auto mode feature replaces repeated manual approval prompts with an automated classifier designed to evaluate tool calls for destructive actions.
- Anthropic stated it will no longer charge for the extra tokens used by the auto mode classifier during tool calls.
- A study of 1,053 paid testers showed that auto mode caught 89% of dangerous commands, whereas human testers caught 13.6%.
- Anthropic reported that Team and Enterprise customers utilizing auto mode ship approximately 25% more pull requests.
Artificial intelligence firm Anthropic has announced a significant update to its Claude Code developer tool, making 'auto mode' the default permission setting beginning August 14. The change applies to Pro, Max, and Team users unless an administrator or individual has explicitly pinned an alternative configuration.
Instead of requiring users to manually approve every tool call, auto mode utilizes an automated classifier to check operations for potentially irreversible, destructive, or out-of-bounds actions. If a command is blocked, the system can attempt a safer route or prompt the user for permission. To facilitate this transition, Anthropic confirmed it will eliminate extra token charges previously applied to the classifier.
To support the safety of the new default, Anthropic cited a study of 1,053 paid testers. The research indicated that human participants caught 13.6% of dangerous commands, a figure that dropped to roughly 5% after 50 prompts, compared to an 89% success rate for auto mode. Despite these findings, Anthropic cautioned that classifiers do not entirely eliminate risk and advised developers to maintain human review for production changes.
The update is intended to make long-running coding sessions more practical. Anthropic noted that Team and Enterprise customers utilizing auto mode currently ship approximately 25% more pull requests than those relying solely on manual approvals.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.