Back to Workshop resources
Claude Cowork2 min read

Is it safe to connect Claude to your work Gmail?

The trust model, the OAuth scopes, what Anthropic does and doesn't do with your data. An honest answer.

Haojun See
Haojun See

Founder & Director, On The Ground

Updated 6 June 2026

Overview

Almost every CW-1 attendee asks this. The answer in two sentences: per Anthropic's Pro plan terms, Claude doesn't train on your chats, and you can revoke OAuth access instantly. The longer answer covers what to actually check before connecting.

What Anthropic does with your data

Per the Pro plan terms: chats are not used for model training. Connectors read data on demand at the time of the request; tokens are stored, not the data itself. Anthropic's enterprise terms (paid Teams and Enterprise tiers) are stricter still. For individuals on Pro, the consumer terms apply.

What to check before connecting

Your employer's AI policy. Many Workspaces explicitly allow, prohibit, or require approval for third-party OAuth. If you're unsure, ask IT before connecting. Your own comfort. Start with read-only and a single connector (Drive scoped to one folder) and expand only after a week of safe use.

How to revoke

Google Account → Security → Third-party apps with account access → Anthropic → Remove. Immediate. No data is left behind on Anthropic's side beyond the conversation transcripts, which you can also delete.

Want to Apply This to Your Business?

We're a Singapore AI development and automation agency. Let's discuss how we can help solve your specific challenges.