Anthropic’s Claude can now send Gmail emails without asking first

Anthropic's AI assistant, Claude, now has the ability to send and reply to emails within Google Workspace without any user intervention or approval. This update to the Claude connector for Google Workspace allows for full Gmail inbox management, including the capability to both send and reply to emails autonomously. The feature also includes the capacity to forward emails without requiring user consent.

While Claude still requires a human user to instruct it to draft or reply to an email, once the response is composed, it can be sent automatically without further user input. However, users can still set controls to require approval before sending an email. By default, Claude asks for approval before sending, but users must modify their settings to permit the AI to bypass this step.

It's important to note that free Claude users cannot utilize this Gmail management feature; they must opt for a paid Claude plan to enable email sending on their behalf.