A company licenses its Google Workspace data for AI training through an export that a super admin starts with Google's Data Export tool, and Google's terms decide which other routes are closed. The export needs a super admin account at least 30 days old and is ready no earlier than 48 hours after it starts, while Google's Workspace API user data policy prohibits using API user data to train or improve an AI model (both checked November 9, 2026).
A Workspace export is one source in AI training data licensing, where a license is one outcome and keeping the archive is another. Not licensing is a valid result.
Each figure below comes from Google's or the program's own page and carries its check date; the DataPacts Editorial Team explains how we source, date and check every figure.
What Exporting Google Workspace Data for a License Involves
Licensing Google Workspace data means granting an AI developer, or a vendor that supplies one, the right to train on an exported copy of the company's Gmail, Drive and Chat records, while the company keeps the original. Google LLC is the vendor named in the Google Workspace terms.
Workspace appears often among the systems that data programs ask for. Mercor's Brendan Foody, who describes himself as its CEO, states that Google Drive appears at 53 % and Gmail at 50 % of the more than 300 enterprises Mercor works with (LinkedIn post, checked November 9, 2026). SimpleClosure, a shutdown service, states that 87 % of the workspaces on its AssetHub include Google Drive (published September 3, 2026, checked November 9, 2026).
Mail as a data type is covered under licensing a company email archive. This page stays on the Google route.
How Google's Data Export Tool Works
Google's Data Export tool copies an organization's Workspace data into a Google Cloud Storage bucket, and a super admin starts it from the Admin console. The table compares the gates and clocks as of November 9, 2026, from Google's help page "Export all your organization's data".
| Gate or clock | Value | Source |
|---|---|---|
| Who starts it | super admin account at least 30 days old | knowledge.workspace.google.com |
| Contact Support first | more than 1,000 users or FedRAMP | knowledge.workspace.google.com |
| Ready | no earlier than 48 hours; typically 72 hours; up to 14 days | knowledge.workspace.google.com |
| Bucket deletion | 60 days from the beginning of the export | knowledge.workspace.google.com |
| Subscription cancelled or suspended | core service data excluded | knowledge.workspace.google.com |
Who Can Start a Data Export and When Support Steps In
Only a super admin whose account is at least 30 days old can start a Data Export, and organizations with more than 1,000 users or FedRAMP authorization contact Google Support first. This page does not list which services or file formats the export covers.
Export Timing and the 60-Day Bucket Deletion
An export is available no earlier than 48 hours after it starts, typically takes 72 hours and can take up to 14 days. Google deletes the Google-provided bucket 60 days from the beginning of the export. The copy for a data room or a buyer's delivery is therefore planned inside that window, counted from the start.
What a Cancelled or Suspended Subscription Changes
Google's help page states that core service data is excluded from the export once the subscription is cancelled or suspended, so the export comes before the account lapses.
The Workspace API User Data Policy and AI Training
Google's Workspace API user data policy governs apps that read Workspace data through Google's APIs, and it prohibits using that data to train AI models.
What the Policy Prohibits for Apps
The policy names two prohibited uses that matter for licensing: transferring or selling user data, and using it to create, train or improve a machine learning or AI model. A company weighing a license places this rule among its other duties in a rights audit for data you want to license. The policy text states that "Transferring, selling, or using user data to create, train, or improve a machine learning or artificial intelligence model is prohibited" (checked November 9, 2026), with an exception for a user's personalized model.
- Transfer or sale of user data obtained through the APIs.
- Use of that data for ML or AI training beyond a user's personalized model.
The policy covers access through Google's APIs. This page does not state what Google's customer terms say about relicensing an export made by the admin tool.
Connector Access by Data Programs Under the Policy
Some data programs connect to a company's systems instead of receiving an export, and a connection that reads Gmail or Drive through Google's APIs runs into the API user data policy. Centific states it connects in-scope systems read-only, Mercor lists a "Connect" step, and Scale states it anonymizes inside the company's environment before extraction (all checked November 9, 2026).
No statement in the sources checked says whether a program's connector is bound by the policy, so a seller asks the program in writing which route it uses. The comparison of buyer connectors vs native exports sets out how the routes differ.
Google's Own Training Commitment vs a Company Licensing Its Workspace
Google's commitment on training its own models with Workspace data is a separate question from a company's decision to license its data to an AI developer. Google states that Workspace does not use customer data to train models without the customer's prior permission or instruction, and refers to a Training Restriction in the Workspace Service Specific Terms (Generative AI in Google Workspace Privacy Hub, checked November 9, 2026). That is Google's claim about its own conduct.
The commitment does not decide whether a company may license its exported records to a third party. User and admin settings for staying out of Google's AI features are covered under opting out of AI training, which this page does not repeat.
Data Programs That Name Gmail, Drive or Google Chat
Ten programs on record name Gmail, Drive or Google Chat among the data they ask for, each listed here with its role and the date its page was checked. The table is alphabetical, carries no payout figures and shows each program's own wording as of November 9, 2026.
| Program | Role | What it states it asks for | One published term relevant to Workspace data | Checked | Source |
|---|---|---|---|---|---|
| Centific Enterprise Data | data vendor | "email, calendar, Teams, Slack, Google Chat"; "Office files, Drive, SharePoint" | read-only connection to in-scope systems; owner picks applications, teams and periods | 2026-11-09 | centific.com/enterprise-data |
| Collinear AI paid data licensing | RL-environment builder | "Gmail" | one-time snapshot; seller keeps ownership | 2026-11-09 | collinear.ai/data-acquisition |
| Handshake AI Data Partnerships | human-data platform | sources named include "Gmail, Google Drive" | "All data is scoped, de-identified, and confidential" | 2026-11-09 | joinhandshake.com/ai/data-partnerships/license |
| Mercor Data Monetization | human-data platform | "Gmail", "Drive" among "100+ supported tools" | Connect step; automated extraction and de-identification | 2026-11-09 | mercor.com/data |
| Nyne data partnerships | data vendor | "chat (Slack, Teams, Google Chat)" | 20+ full-time U.S. W-2 employees, 3+ years | 2026-11-09 | nyne.ai/data-partnerships |
| Ooak Data partner program | RL-environment builder | "Gmail" | raw data deleted within 30 days of delivery | 2026-11-09 | ooakdata.com/partner |
| Polyshares data licensing | data vendor | "Drive and SharePoint" | "defined term" on the home page vs "Perpetual exclusive license" in its track record | 2026-11-09 | polyshares.com |
| Scale Data Partnerships | human-data platform | "chat (Slack, Teams, Google Chat)" | anonymizes inside the company's environment before extraction | 2026-11-09 | scale.com/data-partnership |
| SimpleClosure AssetHub | intermediary (shutdown service) | "WorkspaceData (Slack, Google Drive, email, Notion, Figma, tickets)" | sensitive employee, compensation, health and personal information excluded | 2026-11-09 | simpleclosure.com/product/assethub |
| Turing Project Lazarus | human-data platform | "wikis and docs (Confluence, Notion, Drive)" | non-exclusive; 90-day Turing-only selling window | 2026-11-09 | lazarus.turing.com |
Intermediaries place or refer data and do not acquire it.
Terms differ between programs. Mercor states payout by wire typically within 2 to 4 weeks of the signed agreement. Turing states a 90-day window in which it alone sells the record, and Polyshares shows two wordings on its own pages, "defined term" and "Perpetual exclusive license", so a seller asks which applies. Handshake shows "Earn $100K to $4M+" on one page and an estimator result of "up to $100,000" on another (checked November 9, 2026). Payouts sit with AI data partnership payouts and program terms compared, each dated.
Staff Mail, Shared Files and Outside Senders in a Workspace Export
A Workspace export holds staff mailboxes and files, plus mail and documents from customers and other outside people, and exporting them is not the same as having a basis to license them.
SimpleClosure states that its AssetHub excludes sensitive employee, compensation, health and personal information (checked November 9, 2026). Google's help pages in the record do not say how files shared from other organizations export, so this page does not describe that case.
The steps for staff are set out under employee notice and consent when licensing internal communications.
Benefits, Risks, Downsides and Alternatives of Licensing Workspace Data
Licensing Workspace data can turn years of mail and files into license income, and it also exposes staff and outside correspondence that cannot be recalled once a model has trained on it.
- Benefits:
- A dormant archive can earn license income; SimpleClosure states that startup assets are typically worth tens of thousands to hundreds of thousands of USD (September 3, 2026, an intermediary's range claim, not a price for Workspace data).
- A time-limited, non-exclusive license leaves the company free to use its own copy.
- Risks:
- Labelbox Prism states that models already trained on the data cannot be unlearned (checked November 9, 2026), so a license cannot be reversed in practice.
- Staff and outside mail can carry personal data and confidentiality duties.
- The API route is closed to AI training under Google's policy for apps.
- Downsides:
- The Google-provided bucket is deleted 60 days after the export starts, so delivery work runs on a clock.
- Rights review and de-identification take time before any buyer sees a sample.
- Alternatives:
- Not licensing, and keeping the archive for internal use.
- Licensing Drive documents only, without mail.
- A time-limited, non-exclusive license.
No option is a default; the choice depends on the rights audit and tolerance for an irreversible grant.
Where Google Workspace Fits Among Source Systems and the Seller Process
Google Workspace is one source system among several a company may license, and its export comes after the readiness check and inventory and before de-identification and the choice of buyer. Compare export rights and mechanics by source system, where Slack data export rights by plan and Microsoft Teams data export follow different rules, GitHub repository export rights cover code, and Microsoft 365 exports for data licensing and selling documents, wikis and Drive files cover the neighboring cases. The full sequence to sell or license company data to AI companies runs in this order:
- Data sale readiness
- Data inventory for an AI data sale
- Company data that cannot be licensed
- PII removal and de-identification
- The data manifest
- Choosing a buyer for your data
- AI data licensing agreement
Workspace Data of a Closed or Insolvent Company
A closed company can license its Workspace data only if someone with authority still holds a super admin account before the subscription is cancelled or suspended. Ooak states that wind-down deals close with a director, liquidator or administrator signing (checked November 9, 2026). Read selling data from a shut-down or defunct company for the rest. Deadlines before a shutdown are covered under exporting company data before shutting down.
Google as Workspace Vendor and as a Bidder for Data
Google LLC is the vendor whose rules govern a Workspace export, and separately it holds the successful bid of USD 10,000,000 for Spirit Airlines' de-identified data archive, a sale not approved as of November 9, 2026 (Bankr. S.D.N.Y. 25-11897, Dkt. 1463; hearing November 17, 2026, Dkt. 1779). The sale is covered under Spirit Airlines' data archive sale, and the profile of Google as a data buyer sets out the rest.
Program Pages Named on This Page
The program terms above come from each program's own page, checked November 9, 2026. Profiles follow, alphabetically within each group.
- Hubs: company data partnership programs, who buys AI training data
- Profiles: Centific, Collinear AI, Handshake AI data buyer profile, Mercor data buyer profile, Nyne, Ooak Data, Polyshares, Scale AI, SimpleClosure (intermediary), Turing Project Lazarus profile
Frequently Asked Questions
Can a company license its Gmail and Drive data to AI companies?
Yes, a company can license an exported copy of its Gmail and Drive data where its contracts and the law allow, usually as a license rather than a sale. The export comes from a super admin, and Google's API user data policy bars apps from training AI models on API user data.
Who can export all of a company's Google Workspace data?
A super admin whose account is at least 30 days old, with Google Support involved first above 1,000 users or under FedRAMP. The rule comes from Google's help page, checked November 9, 2026.
How long does a Google Workspace data export take?
Google states an export is ready no earlier than 48 hours, typically after 72 hours and at most after 14 days, and the bucket is deleted 60 days from the start. The figures are from Google's help page, checked November 9, 2026.
Does Google train AI models on Workspace data?
Google states that Workspace does not use customer data to train models without the customer's prior permission or instruction. That is Google's claim, checked November 9, 2026.
What is Google Workspace data worth to an AI buyer?
No public price exists for Workspace data as such; SimpleClosure, an intermediary, states startup assets are typically worth tens of thousands to hundreds of thousands of USD (September 3, 2026). Methods are in how much company data is worth to AI companies.