Anthropic's Claude Gains Auto Mode for Coding

Anthropic has introduced a new 'auto mode' for its AI assistant Claude, enhancing its coding capabilities with increased autonomy. This feature, currently in a research preview phase, aims to reduce the need for developers to manually approve every AI-driven action.

Auto mode evaluates each proposed action for potential hazards, such as unintended behaviors or prompt injection attacks, before execution. Actions deemed safe are performed automatically, while those flagged as risky are prevented. This advancement builds on Anthropic's previous tools for automating development workflows.

The company plans to make auto mode available to enterprise and API users in the near future. For now, it is restricted to certain model versions and recommended for use in controlled environments during its testing phase.

This development aligns with a broader trend in AI towards greater independence in task execution. However, ensuring that enhanced autonomy does not compromise reliability and security remains a key challenge for the industry.