MUMBAI: Claude is being given a little more freedom to code. Anthropic is making Auto Mode the default setting for most paid Claude Code users from 14 August, allowing the AI coding assistant to carry out more programming tasks without repeatedly asking for permission. The change will apply to Pro, Max and Team subscribers, while Enterprise users will continue to have Auto Mode as an opt-in feature for now. The move marks a shift from Claude Code’s current workflow, where the assistant frequently pauses before executing commands, editing files or taking other programming actions.
Under Auto Mode, Claude will continue working independently unless an action is considered potentially irreversible, destructive or outside the user’s authorised environment. A safety classifier will assess individual actions and decide whether Claude can proceed without human approval.
If a command is deemed risky, Claude can either find a safer alternative or ask the developer for permission. If it repeatedly encounters safety blocks, the system can automatically return to manual approval mode.
Anthropic said the decision follows internal testing as well as a study involving 1,053 paid users. The company found that human reviewers identified and rejected dangerous commands only 13.6 per cent of the time, while Auto Mode blocked 89 per cent of the same harmful actions.
The findings also showed that users approved around 97 per cent of permission prompts, suggesting that repeated safety requests could become little more than a click-through exercise. Anthropic argues that reducing routine prompts could make the warnings that remain more meaningful to developers.
There is also a productivity angle. Auto Mode is designed to allow Claude Code to work for longer periods on large and complex coding projects without constant interruptions. Anthropic said the feature is particularly relevant to newer models such as Claude Opus 5, which are built for extended tasks.
Teams and Enterprise customers using Auto Mode have reportedly generated around 25 per cent more pull requests, according to Anthropic. Companies including Adobe, Nuro, Gusto and Garner Health are already using Auto Mode as their production default.
The extra autonomy does come with a bigger security question, however. Anthropic said it has added protections including prompt-injection screening, configurable deny rules and safeguards intended to stop sensitive information from being moved outside authorised environments.
Users who have already selected a different default permission mode will receive a one-time notification and can continue switching between settings.
For developers, the change essentially turns Claude Code from an assistant that frequently asks “May I?” into one expected to keep moving unless something looks dangerous. Anthropic says Auto Mode has already been used internally for months, with the company using real-world testing to identify and block potentially harmful actions.