The Top 10 SaaS Tools Using Your Data for AI Training (2025 List)

What they do: After May 2024 controversy, Slack clarified they do NOT use customer data to train generative AI. They only use data for non-generative features (emoji suggestions, search).

The policy: "Slack will not use Customer Data to train generative AI models unless Customer provides affirmative opt-in consent."

Opt-out: Generative AI training requires opt-in. For non-generative features, contact [email protected] to opt out.

What happened: June 2024 controversy over vague TOS language caused massive backlash. Adobe reversed course and clarified their policy.

Current policy (June 24, 2024): "Adobe will not use your Local or Cloud Content to train generative AI" (except Adobe Stock submissions).

Opt-out: Not needed. Updated TOS explicitly excludes AI training on customer content.

What they do: After August 2023 backlash, Zoom clarified they do NOT use customer content (audio, video, chat) for AI training. Only uses telemetry/diagnostic data.

Current policy: "Zoom does not use audio, video, chat, screen-sharing, attachments or other communications-like Customer Content to train AI models."

Opt-out: Account admins can disable AI Companion features in Account Settings → AI Companion.

What they do: GitHub does NOT train on private repos (Business/Enterprise). BUT: 2025 issue - temporarily public repos cached by Bing remain accessible via Copilot even after made private.

The risk: 20,000+ private repos from 16,000 organizations still exposed through cached "zombie data" in Microsoft Copilot.

Mitigation: Use Business/Enterprise plans. Never make sensitive repos public, even temporarily.

What they do: CLASS ACTION LAWSUIT (Nov 2025): Figma accused of secretly harvesting user designs, layers, text, and images to train AI without consent.

The issue: Free/Pro users automatically opted IN to AI training by default. Only Enterprise/Org users opted OUT by default.

Opt-out: Admins can disable in team settings, but many users didn't know they were opted in.

What they do: Strong privacy policy. Notion explicitly does NOT train AI models on workspace content. Zero data retention for Enterprise plans.

Current policy: "Your use of Notion AI does not grant us any right to train ML models on your Customer Data."

Data retention: Zero retention (Enterprise), max 30 days (other plans). OpenAI embeddings don't retain data.

What they do: Dropbox may use file metadata and interaction data for AI features, but not file content (as stated).

The clause: "We use metadata to improve AI-powered search and organization features."

Opt-out: Limited; contact support for enterprise options.

What they do: Google explicitly does NOT use customer data (including prompts) to train Gemini models for paid Workspace customers.

Current policy: "Your data is not used to train Gemini models or for ads targeting. Content is not human reviewed or used for model training."

Exception: Free Gemini Enterprise Starter (after trial) DOES use data for training unless you opt out in settings.

What they do: Microsoft explicitly does NOT use customer data from Microsoft 365 to train foundation LLMs. This applies to both commercial AND personal/family subscriptions.

Current policy (Nov 2024): "Prompts, responses, and Microsoft Graph data aren't used to train foundation LLMs, including those used by Microsoft 365 Copilot."

Note: Microsoft DOES use data from Bing/MSN for AI training, but keeps M365 data separate.

What they do: Canva's AI features may use your designs to improve templates and suggestions, though they claim private designs are protected.

The clause: "We may use anonymized design data to improve AI recommendations."

Opt-out: Review AI feature settings in your account.