Does Slack AI train on your data?
Slack (Salesforce) · Workplace AI suite · official site ↗
Doesn’t train on your dataZDR / DPA on enterpriseNo training to opt out of
Does Slack AI train its models on your data?
No — Slack AI does not use your content to train its AI models by default.
Slack does not use customer data to train large language models or any generative AI, and customer data never leaves Slack's trust boundary; Slack AI uses retrieval (RAG) at inference time instead.
Can you opt out?
No opt-out needed for generative AI — Slack will not use customer data to train generative models without affirmative opt-in. Note: Slack's older, non-generative ML features (e.g. search ranking, emoji/channel suggestions) historically run on an opt-out basis; an org can email Slack to opt out of those.
Zero retention / DPA
Slack AI is built so customer data stays within Slack's trust boundary and is not used to train third-party LLMs; data access is scoped to what each member can already see. Enterprise plans add admin and compliance controls. DPA ↗
What the listicles get wrong
Distinguish the two: Slack AI (generative) does not train on your data, but Slack's older non-generative ML has been opt-out — the source of the 2024 backlash.
Verdict by plan tier
Slack AI (paid add-on, all plans)No trainingNo customer data used to train Slack's or third-party LLMs; RAG only; results limited to data the requesting member can access.
Global (non-generative) ML modelsTrains unless you opt outTraditional ML for search/suggestions historically uses de-identified, aggregated data on an opt-out basis; org owners can email to opt out.
Get notified when this changes
We track Slack AI's data-training and retention policy. Leave your email and we'll send one note if it changes.
Frequently asked questions
Does Slack AI train its AI models on my data?
No — Slack AI does not use your content to train its AI models by default. Slack does not use customer data to train large language models or any generative AI, and customer data never leaves Slack's trust boundary; Slack AI uses retrieval (RAG) at inference time instead.
Can I opt out of Slack AI training on my data?
There is no training opt-out to set for Slack AI: No opt-out needed for generative AI — Slack will not use customer data to train generative models without affirmative opt-in. Note: Slack's older, non-generative ML features (e.g. search ranking, emoji/channel suggestions) historically run on an opt-out basis; an org can email Slack to opt out of those.
Does Slack AI offer zero data retention (ZDR) or a DPA?
Slack AI is built so customer data stays within Slack's trust boundary and is not used to train third-party LLMs; data access is scoped to what each member can already see. Enterprise plans add admin and compliance controls.
Is Slack AI safe to use with confidential or proprietary data?
It depends on your plan tier. Slack AI (paid add-on, all plans): No customer data used to train Slack's or third-party LLMs; RAG only; results limited to data the requesting member can access. Global (non-generative) ML models: Traditional ML for search/suggestions historically uses de-identified, aggregated data on an opt-out basis; org owners can email to opt out. Always confirm current terms with Slack (Salesforce) before sending confidential data — this is cited public information, not legal advice.
Sources
https://slack.com/trust/data-management/privacy-principles
https://slack.com/blog/news/how-we-built-slack-ai-to-be-secure-and-private
https://www.ciodive.com/news/slack-ai-policy-language-data-privacy/716687/
This page is cited public information, not legal or compliance advice. Whether Slack AI trains on your data, and any zero-retention or DPA option, can depend on your specific plan, region and contract. Always confirm current terms with Slack (Salesforce) before sending confidential or proprietary data.