Gmail: Claude can now click “Send” without asking for your permission.

Caught up in a minor controversy following its decision to watermark Claude’s texts, Anthropic is making progress in other areas.Claude is enhancing its Gmail integration, and the AI assistant can now reply to, forward, and send emails directly from your inbox without requiring your approval for every send.

From the assistant that writes to the assistant that acts

The Google Workspace connector has been around for some time now, allowing Claude to interact with Gmail, Google Calendar, and Google Drive. Until now, the AI could at most prepare a response that the user then had to manually copy and paste into Gmail. This manual step is now gone.

Claude can now send emails in Gmail and manage files in Google Drive.

Ask Claude to reply to a thread, and it drafts and sends the response. You control when it needs your approval.

Connect Gmail or Google Drive from the connectors menu to try. Available on all paid plans. pic.twitter.com/cFZEjh3MgB

— Claude (@claudeai) August 18, 2026

Now all you have to do is ask Claude to reply to an email, and the assistant will analyze the conversation thread, draft a context-appropriate response, and send it directly. Claude can also forward messages to a third party.

Approval enabled by default, but can be disabled

By default, Claude still asks for confirmation before each send, allowing you to review the text and verify the recipient before the message is actually sent.This validation step can, however, be disabled by the user, who then allows the assistant to send its messages without repeatedly asking for confirmation each time. On Team and Enterprise plans, organization administrators decide whether members of their team can enable this enhanced autonomy.

This feature remains exclusive to Claude's paid plans.

Practical, but not without risk

The time savings are obvious for repetitive tasks: follow-ups, appointment confirmations, routine responses to customers. However, the shift from an assistant that suggests to an assistant that acts raises a question asked by Digital Trends: what happens if Claude misinterprets the context of a conversation? A poorly worded response is easy to correct as long as it remains in the drafting window. Once sent, the error becomes much harder to remedy. The validation system therefore remains the main safeguard, provided it is not disabled carelessly.

This new development fits into a broader approach for Claude, which can already also manage events in Google Agenda and perform various actions in Google Drive, with the aim of managing its email and organization through simple natural-language instructions rather than juggling between several interfaces.

👉🏻 Follow tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.

Source:9to5Google