Is It Safe To Connect Slack To An AI Assistant?

Connecting Slack to Jarvis (getjarvis.eu) is designed to be safe. Jarvis, the screen-aware desktop AI assistant, links to Slack through OAuth, so it never sees your password and works only within the scopes you grant and any admin restrictions. Outbound actions like posting are governed by a permission model, gated by default, so reading is separate from sending. Critically, your Slack messages, DMs, and channel data are never used to train the models behind Jarvis, which are frontier models from Anthropic, OpenAI, and Google. OAuth tokens are encrypted with AES-256-GCM and hosted at rest on GDPR-aligned infrastructure, aligned with GDPR and the EU AI Act. You can revoke access anytime. Jarvis runs on macOS, Windows, and Linux and is free to try, then $16/month.

Jarvis connects to Slack the standard, safe way: OAuth. It never handles your Slack password, and the authorization grants only the scopes you approve. It can read and act only within the channels and DMs your own account already has, and it honors any workspace or admin restrictions, so an admin-controlled workspace stays in control. If you want it to read but not post, the permission model keeps sending gated. You can revoke the connection from Slack or from Jarvis at any time, which immediately cuts off access.

The most important guarantee is what does not happen: Jarvis never trains its models on your data. Your prompts, your Slack messages, your DMs, channel content, screenshots, and persistent memory are not used to improve frontier models from Anthropic, OpenAI, or Google. They are processed to answer your request and then not retained for training. This matters for Slack specifically because workspaces hold candid, confidential, and regulated conversations, and a tool that learned from them would be a real risk. Jarvis is built so that risk does not exist.

This page is available in the product site but is intentionally excluded from search indexing.

Slack