For developers using Anthropic’s Claude, the latest update brings a welcome change. Gone are the days of babysitting every action; now, Claude can decide which actions to take on its own—though with some limits.

This shift towards autonomous coding tools is part of a broader trend in the tech industry. Companies like GitHub and OpenAI have introduced similar features, but Anthropic’s Auto Mode takes it a step further by allowing the AI to make decisions about when to ask for permission.

The safety layer included with this auto mode reviews each action before it runs, checking for risky behavior or hidden malicious instructions. Safe actions proceed automatically while risky ones are blocked. This feature builds on Claude Code Review and Dispatch for Cowork, tools designed to catch bugs and manage tasks respectively.

However, developers will need more information about the specific criteria used by Anthropic’s safety layer before fully embracing this new mode. As AI tools become more autonomous, questions of trust and control are bound to arise.