Anthropic is set to make Auto Mode the default setting in Claude Code for Pro, Max and Team accounts from August 14, 2026, reducing the need for developers to manually approve every action performed by the AI coding assistant.
The move marks another step towards more autonomous AI-powered software development, as coding agents increasingly take on multi-step programming tasks with limited human intervention.
Claude Code to Require Less Human Approval
Claude Code currently asks users for permission before carrying out certain actions. Under Auto Mode, the system will be able to proceed without requesting approval at every stage.
Anthropic said Claude Code will continue to pause when an action is considered:
- Irreversible
- Destructive
- Outside the user’s environment
The company introduced Auto Mode experimentally in March as an attempt to balance the speed of autonomous coding with appropriate safety controls.
With the feature becoming the default, developers will be able to allow Claude Code to complete longer and more complex programming workflows without repeatedly responding to permission prompts.
Anthropic Says Auto Mode Can Be Safer
Interestingly, Anthropic’s testing suggests that giving the AI greater autonomy may actually result in better safety outcomes than relying entirely on users to manually approve actions.
In a study involving 1,053 paid Claude Code testers, Anthropic said Auto Mode detected 89% of harmful actions, compared with only 13.6% detected through human review.
The company attributed part of this difference to the behaviour of users when repeatedly presented with permission requests.
According to Anthropic, users approved around 97% of permission prompts in Claude Code, suggesting that frequent approval requests can eventually become routine rather than meaningful security checks.
Developers Are Already Using Auto Mode
Claude Code Head Boris Cherny said that he and his team have been using Auto Mode exclusively for months.
The comment highlights Anthropic’s confidence in moving away from a workflow where developers continuously monitor and approve individual actions taken by the coding agent.
For developers working on large projects, the change could make Claude Code feel more like an autonomous software-engineering assistant rather than a traditional AI chatbot that waits for instructions after every step.
New Safety Features Added
Anthropic said it is also strengthening Claude Code’s safety systems as it increases the level of autonomy available to the agent.
Among the safeguards being introduced are prompt-injection screening and customisable hard-deny rules.
These controls are designed to prevent potentially dangerous behaviour, including attempts to exfiltrate sensitive data.
Prompt injection is a particularly important concern for autonomous coding agents because they can access files, execute commands and interact with development environments. Malicious instructions hidden inside code, documentation or other files could potentially influence an AI agent’s behaviour.
AI Coding Moves Towards Greater Autonomy
The move reflects a broader shift in the AI coding industry.
Tools such as Claude Code and competing AI coding agents are increasingly being designed to handle entire development workflows—from understanding a codebase and writing code to testing, debugging and making changes across multiple files.
Making Auto Mode the default could further accelerate this transition by allowing developers to delegate longer sequences of tasks to AI agents.
At the same time, greater autonomy increases the importance of strong safeguards because an AI agent capable of independently executing commands can potentially make mistakes at a much larger scale.
Anthropic’s approach is therefore focused on allowing Claude Code to operate independently for routine tasks while maintaining restrictions around actions that could cause irreversible damage, leave the user’s environment or expose sensitive information.
Starting August 14, Auto Mode will become the default for eligible Pro, Max and Team accounts, marking another step towards more autonomous AI-assisted software development.
Author
About the Author
