Can Claude or ChatGPT send emails?
Short answer: yes. Claude and ChatGPT can both send email for you now, straight from your own Gmail or Outlook. And both ask you first.
But “send this email” and “send my newsletter” are very different jobs. Here is what each AI can do today, where it stops, and how to let AI help with your newsletter without the scary part.
What Claude can do
Connect Gmail in Claude’s settings, under Customize and then Connectors. After that, Claude can read your email, write drafts, and send, reply and forward.
Claude asks before every send. On team plans, the owner can change that.
Outlook works too. With a personal Outlook or Hotmail account, Claude can write and send, and it asks first. With a work account, your admin has to allow it.
You may have read that Claude can only write drafts. That was true for a while. Anthropic’s help pages now say it can send.
What ChatGPT can do
Since June 2026, ChatGPT can send email from a chat on most paid plans. Connect Gmail or Outlook in its settings. ChatGPT writes the draft, and you choose to send it.
You also decide how much it may do without asking, from “always ask” to letting it act on its own.
Why a newsletter is a different job
Gmail and Outlook are made for one person writing to a few people. A newsletter comes with rules they were never built for:
- Limits. Gmail can stop you sending for up to a day if one email goes to more than 500 people.
- Unsubscribe. Google and Yahoo want a one-click unsubscribe in every marketing email.
- Spam reports. If too many people mark you as spam, Gmail starts sending you to the spam folder.
- The law. In the US, every marketing email needs your postal address and a clear way to opt out.
A newsletter tool takes care of all of that. It keeps your list, adds the unsubscribe link, stops emailing people who leave, and sends through a service built for big lists.
So the safe setup is simple: your AI talks to your newsletter tool, and the newsletter tool does the sending.
The two real risks
First, email can’t be unsent. One wrong send to 5,000 people stays sent.
Second, AI can be tricked. Hidden text in a web page or an email can tell it to do something you never asked for. OpenAI and Anthropic both warn about this. It’s called prompt injection.
What keeps you safe is the same in every tool. Let the AI write drafts and send the test to you. Read it. Then press send yourself. And only email people who asked to hear from you.
How Mailpuff lets AI help
Mailpuff is a newsletter plugin for WordPress, and it works this way out of the box.
Connect Claude, ChatGPT, Cursor or Codex to your WordPress, then ask in plain words:
- “Write a newsletter about our spring sale and send me a test.”
- “How did last Tuesday’s email do?”
There is one switch, called “AI agents can send campaigns”, and it is off until you turn it on. While it’s off, your AI can write and edit drafts, but it has no way to send or schedule a campaign. Tests go only to you. And your AI can’t flip the switch itself. Only you can, in Mailpuff’s settings.
So when the test lands in your inbox, you read it, and you press send. We think that last step should stay yours.
Setup takes a few minutes. The guides are Connect an AI agent to Mailpuff and AI agent permissions and safe sending. If you are also wondering what the sending itself costs, see what it costs to send a newsletter.
Sources
All checked on 5 November 2026.
- Anthropic, Use Google Workspace connectors: https://support.claude.com/en/articles/10166901
- Anthropic, Connect to Microsoft 365: https://support.claude.com/en/articles/15183774
- OpenAI, ChatGPT release notes (8 June 2026): https://help.openai.com/en/articles/6825453-chatgpt-release-notes
- OpenAI, Managing app permissions in ChatGPT: https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt
- OpenAI, Connected apps in ChatGPT: https://help.openai.com/en/articles/11487775-connected-apps-in-chatgpt
- Google, Limits for sending and getting mail: https://support.google.com/mail/answer/22839
- Google, Email sender guidelines: https://support.google.com/a/answer/81126
- Yahoo, Sender best practices: https://senders.yahooinc.com/best-practices/
- FTC, CAN-SPAM Act compliance guide: https://www.ftc.gov/business-guidance/resources/can-spam-act-compliance-guide-business
- PlugPress, Connect an AI agent to Mailpuff with MCP: https://plugpress.io/docs/mailpuff-mcp-setup/
- PlugPress, AI agent permissions and safe sending: https://plugpress.io/docs/mailpuff-ai-agent-safety/