Anthropic to Make Claude Code’s Auto Mode the Default
Anthropic says Claude Code’s automatic mode will reduce approval prompts while using AI-powered safety checks to identify potentially harmful or destructive actions.
Anthropic is preparing to make Auto Mode the default setting for Claude Code on Pro, Max and Team accounts starting August 14, 2026. The change is designed to make AI-assisted programming more autonomous by reducing the need for developers to manually approve individual actions taken by Claude Code.
Under the new system, Claude Code will be able to carry out many programming tasks automatically rather than stopping to request permission at every step. However, Anthropic says safeguards will remain in place for actions considered irreversible, destructive or directed toward systems outside the user’s environment. This approach is intended to give developers greater speed while maintaining a layer of automated oversight.
Anthropic’s decision is based partly on testing that suggested automated safety checks could outperform humans when reviewing large numbers of AI-generated actions. The company reported that its classifier identified 89% of harmful actions in testing, compared with 13.6% detected by human reviewers. Anthropic argues that people can become less attentive when repeatedly asked to approve similar prompts, creating what it describes as approval fatigue.
The change represents a significant step toward more autonomous AI coding agents. Claude Code can already work with files, run commands and interact with development environments, meaning greater autonomy could allow developers to delegate larger portions of software-development workflows to the AI. Anthropic has previously emphasized that as AI agents become more capable, controlling their potential “blast radius” becomes an important engineering challenge.
For developers, the update could make Claude Code faster and less interruptive, but it also increases the importance of understanding how the system handles permissions and risky actions. Anthropic is effectively shifting some responsibility from constant human approval toward automated safety systems. If the approach works as intended, it could become an important model for how future AI agents balance autonomy, productivity and security.
