Slack AI training opt out: the exact email, and who has to send it
The Morse teamPublished 9 min read
The Slack AI training opt out is an email, and only an owner can send it. As of October 24, 2026, Slack's privacy principles say an Org Owner, Workspace Owner, or Primary Owner has to write to feedback@slack.com with the workspace or org URL and the subject line "Slack Global model opt-out request" (Slack). There's no switch in Slack's settings for it, and a regular member can't opt out alone. Until an owner sends that email, Slack's predictive models learn from your workspace by default.
We read Slack's privacy principles, its AI principles, its help pages on AI security, AI feature access, and data retention, its supplemental terms, its API terms, and its privacy policy on October 24, 2026. Slack deserves credit for being specific: it names the models it trains, gives examples of each, and says plainly which AI features keep data and for how long. Two kinds of AI run in Slack, and each has its own control.
Does Slack train AI on your messages?
Yes, for its predictive models, unless your workspace opts out. Slack's privacy principles say that "to develop predictive models for features such as emoji and channel recommendations, our systems analyze Customer Data (e.g. messages, content, and files)" (Slack). Slack calls these global models. Its examples are channel recommendations, search ranking, autocomplete, and emoji suggestions, and it says it doesn't build them "in such a way that they could reproduce some part of Customer Data."
Slack's post explaining the policy says "workspaces are not opted out by default" (Slack). The contract says the same thing in legal terms. Slack's supplemental terms, last updated February 27, 2026, let Salesforce process Customer Data "to train models for use by the services and features that Customer has access to" (Slack).
For the generative side, the answer is no, by Slack's account. The AI principles page says your Customer Data "(like messages and files) is never used to train any LLMs" (Slack). Slack's AI security page adds that the LLMs run inside Slack's own cloud environment and "do not retain information once the request has been processed" (Slack).
The two pages word that promise differently. The privacy principles page says Slack "will not use Customer Data to train generative AI models unless Customer provides affirmative opt-in consent" (Slack). So the LLM promise has an opt-in exception on one page and none on the other. Neither page describes how a customer would give that consent.
What does the Slack opt-out cover, and what doesn't it?
The email covers only the global predictive models. Here is every AI and data control we found on Slack's pages, who can use it, and what it changes.
Read October 24, 2026.
| Control | What it changes | Who can use it | Where | Source |
|---|---|---|---|---|
| Global model opt-out | Your workspace's data stops training Slack's shared predictive models; it still improves your own workspace | Org Owner, Workspace Owner, or Primary Owner | Email to feedback@slack.com | Privacy principles |
| AI feature access | Which AI features (summaries, huddle notes, search answers, recaps, Slackbot) anyone can use | Owners and admins | Admin, Workspace settings, Roles & permissions, Feature access, AI | Manage access to AI features |
| Member feedback on AI responses | Whether members can send Slack feedback on AI answers; blocked by default in some workspaces | Owners and admins | Same AI feature access screen | Manage access to AI features |
| Data retention | How long Slack keeps messages, files, canvases, and lists before deleting them | Workspace Owners, or Org Owners by policy | Admin, Workspace settings, Data retention | Customize data retention |
| Member retention overrides | Lets members set their own retention for private channels and their DMs, on paid plans | Members, if an owner allows it | Conversation settings | Customize data retention |
| Third-party apps | Apps offered outside your organization may not train an LLM on API data | Set by Slack's API terms | None to change | API terms |
Opting out by email doesn't turn off Slack's AI features, and turning off AI features doesn't opt you out of global model training. Several guides that rank for this query blur the two. One says the email is "only guaranteed to stop Slack from sharing your personal data with AI third-party companies for training" (AgainstData), which doesn't match Slack's pages: the opt-out is about Slack's own predictive models, and Slack says LLM providers "don't have access to Customer Data" at all (Slack). Others say an owner "or administrator" can send it, while Slack's page names only owners.
How do you opt out of Slack AI training?
If you're an owner, it takes one email. Slack's privacy principles give the format (Slack):
- Check your role. The request has to come from an Org Owner, Workspace Owner, or Primary Owner. Admins aren't on Slack's list.
- Write to feedback@slack.com. Slack's page doesn't say which address the email must come from. We'd send it from the email on the owner's Slack account, so Slack can match it.
- Use the subject line exactly: "Slack Global model opt-out request".
- Put your Workspace or Org URL in the body, for example yourteam.slack.com. On an Enterprise org, send the org URL.
- Keep Slack's reply. Slack says "we will process your request and respond once the opt out has been completed." Until that reply arrives, assume you're still opted in.
After that, Slack says your Customer Data "will only be used to improve the experience on your own workspace," and you keep the benefits of its global models "without contributing to the underlying models."
What can you do if you're not a Slack owner?
Ask an owner, and give them the request ready to send. Slack's privacy policy says the organization that signed the contract "controls its instance of the Services" and its data, and tells members with questions about workspace privacy to "contact the Customer whose Workspace you use" (Slack). The same policy says you can find your owners' and admins' contact details at slack.com/account/team while signed in.
A message you can copy:
Hi, could you opt our Slack workspace out of Slack's global model training? Slack's privacy principles say it has to come from an owner: email feedback@slack.com from your Slack account, with the subject "Slack Global model opt-out request" and our workspace URL in the body. Slack replies once it's done. Details: https://slack.com/trust/data-management/privacy-principles
While you wait, you have two controls of your own, if your workspace allows them. On paid plans, an owner can let members set their own retention for private channels and DMs (Slack), so those messages are deleted sooner. And Slack's AI features only use content you can already see, so they "won't display or use data from private channels or direct messages (DMs) they aren't a part of" (Slack). The other people in a DM can still see it, though.
How do you turn off Slack AI features?
Owners and admins do it per feature. On Pro, Business+, and Enterprise Select, click Admin in the sidebar, choose Workspace settings, then Roles & permissions, Feature access, and AI. Next to each feature, click Edit and choose Everyone or No one (Slack). On Enterprise Grid and Enterprise+, the same choice sits under Organization settings, and you can also limit a feature to specific people and groups.
Slack's plans page explains why more workspaces now have these features. In June 2025, Slack folded AI into its paid plans and stopped selling the Slack AI add-on on its website, with existing add-on buyers moved over at their first renewal after August 17, 2025 (Slack).
How long does Slack keep your messages and AI summaries?
As long as your workspace's retention setting says. On paid plans, Slack keeps data "for the lifetime of your workspace by default." On the free plan, owners choose 90 days or one year, and older data is deleted on a rolling basis (Slack). Deletions run once a day and can't be undone.
The AI features have their own periods, all on Slack's AI security page (Slack):
- Conversation summaries and search answers are ephemeral. Slack says the data "is not stored on devices or servers."
- Recaps are stored for 90 days, and deleted early if the messages they summarize are deleted.
- Channel summaries from a workflow become ordinary messages or canvases, kept under your retention settings.
What happened with Slack's privacy principles in 2024?
People noticed the training clause in May 2024, and Slack rewrote the page. Slack's post on the change says it "heard from some in the Slack community that our published privacy principles weren't clear enough," and that "we realised that they were right" (Slack). The post says Slack's traditional models "use de-identified, aggregate data and do not access message content in DMs, private channels or public channels," and that opting out removes some feedback prompts but turns off no features.
That post is hard to find now. On October 24, 2026, its US address loaded a different Slack article about machine learning, and the UK copy we cite is dated April 7, 2025. The opt-out itself hasn't changed: the same email, the same subject line, owners only.
How Morse keeps your questions out of anyone's training
In Slack, the customer is your employer, and every control in the table above belongs to them. You can't opt yourself out, and your messages stay on Slack's servers for as long as the owner's retention setting says. That's how workplace software is meant to work, but it means the questions you'd rather keep to yourself shouldn't go into a work tool you don't control.
Morse puts the controls with you. Morse doesn't train AI on your content, sell your data, or show ads, so there's no email to send and no owner to ask. Your chat history is stored in your browser, encrypted, not on our servers (privacy policy), so there's no workspace retention setting deciding how long it lives. Each chat has a privacy mode:
- Private, for drafting the question before it goes to your team. It's the default for a new chat. Open-weight models like DeepSeek, Kimi, and GLM, plus Grok, run under Morse's zero data retention agreements with its inference, search, and other providers, so they keep nothing. Where a provider confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't. We explain the term in what zero data retention means.
- E2EE, for anything no one should read. Your browser checks the enclave's hardware attestation before anything leaves your device, then encrypts the message so only that enclave can read it. Morse's servers relay text they can't read. This rests on the enclave hardware and attestation working as designed.
- Anonymous, for GPT, Claude, and Gemini without the provider knowing it's you. Our proxy removes your account, email, and IP address and sends the message under Morse's own account. The provider still sees the words and may keep them for a limited time to check for abuse, under its own policy, so leave out names and details that point to you.
In every mode, the history locks when you sign out and takes two keys to open, one of which never leaves your device. Clearing your browser or deleting your account removes your chats for good, and Morse can't recover them, so keep a copy of anything you need. Because we don't store chats, we usually have very little to give authorities.
So what should you do about Slack AI training?
- You're an owner: send the email above and keep Slack's reply. If you also want AI features off, change them under Feature access, since that's a separate control.
- You're a member: send your owner the message above, and ask whether members can set their own retention for DMs.
- You want answers no employer or app trains on: ask them somewhere you hold the account. On Morse, there's nothing to opt out of.
For how 15 other AI apps handle training, see which AI companies train on your data, and for Microsoft's workplace assistant, is Copilot private.
Try a chat from the Morse home page without an account, or start free with no card. End-to-end encrypted chats come with Pro, at $19 a month.
Common questions
Does opting out turn off emoji or channel suggestions?
No. Slack's clarification post says no product features are turned off, though some places where users could give feedback are removed, and the global models will "likely perform slightly worse" on your team.
Do third-party Slack apps train AI on my messages?
Slack's API terms, effective October 10, 2025, say an app offered outside your organization may not "use API Data to train a large language model." An app your own company builds for internal use falls outside that rule, so ask your admin what's installed.
Can my Slack owner or admin see my direct messages?
Sometimes. Slack's privacy policy says owners and admins "may be able to access, modify, or restrict access to Information," and its retention help page describes workspaces approved to export data from all conversations, DMs included.
Does Slack AI cost extra?
Not since 2025. Slack's plans page says it stopped selling the Slack AI add-on on its website, and conversation summaries and huddle notes now come with every paid plan, while recaps, translations, and search answers need Business+ or Enterprise+.