Software and SaaS companies
GitHub, Slack, Jira and Linear data export: what the terms say about ownership, AI use and export
On this page
The short answer
GitHub, Atlassian (Jira and Confluence) and Linear terms all say your company owns its content, so data export for an AI license is mostly a question of method and timing. Slack's API terms bar outside apps from using API data to train a large language model, so export Slack data yourself (source: Slack API Terms of Service (effective Oct 10, 2025)); private channels need Business+ or Enterprise (source: Slack Help Center, export your workspace data).
Each export misses something. GitHub migrations skip Actions run history, Linear CSVs skip attachments, and Jira backups skip automation flows. Retention after cancellation runs 30 to 60 days at most vendors, so export before you cancel.
Why your own export comes first
Some AI data programs offer connector apps that pull directly from your tools, and some say they cover extraction at no cost to the seller (source: program terms, Regulr buyer directory). Each figure or claim about programs on this page is a program's own advertised term; our buyer directory lists the programs and links to their published terms.
Connectors are convenient, but a vendor's API terms govern what an outside app may do with the data it pulls, and those terms are increasingly written with AI training in mind. Slack's are the clearest example (below). An export made by your own admin, using the vendor's built-in tools, is your company's copy of its own records. In our view it is the cleaner starting point for any license, and it gives you a file you can review before a buyer sees anything.
Two questions for every system
What does this export include, and how long will the vendor keep our data if the subscription ends? The answers below come from each vendor's own terms and documentation, read on October 18, 2026. Your signed order form or enterprise agreement governs if it says something different.
GitHub terms: who owns your code, and does GitHub train on private repos
What the terms say
- Ownership. Section D.3 of GitHub's Terms of Service says “You own Your Content.” (source: GitHub Terms of Service (effective Apr 27, 2026))
- GitHub's license. Section D.4 grants GitHub and its affiliates the right to store, host, parse and copy your content as necessary to provide, develop and improve the service, “including by training AI Features” (source: GitHub Terms of Service (effective Apr 27, 2026)). The license is nonexclusive. It does not stop you from licensing your own copy, but a buyer asking for exclusive AI training rights should know it exists.
- Private repositories. Section E.2 says GitHub “considers the contents of private repositories to be confidential to you.” Section E.3 says that if you provide private repository content as input to AI features, GitHub may use that input to train and improve the service, subject to the opt-out in Section J.3, and that it will not otherwise use private repository contents to develop or improve the service (source: GitHub Terms of Service (effective Apr 27, 2026)).
- Business accounts. Section J says the training provisions in J.3 apply only to individual licenses; use under a GitHub Customer Agreement or volume licensing agreement is governed by that agreement (source: GitHub Terms of Service (effective Apr 27, 2026)).
How to export a GitHub repository with full history
Git itself is the easy part: a mirror clone copies every branch, tag and commit. The records around the code are what get lost. GitHub's migration documentation lists what a repository migration carries, including git source with commit history, pull requests, issues, milestones, wikis, Actions workflows and commit comments, and what it does not, including Actions workflow run history, commit status checks, the edit history of issue and pull request comments, code scanning results and Discussions (source: GitHub Docs, about migrations between GitHub products).
CI history has a shorter clock than most owners expect. GitHub keeps Actions artifacts and logs for 90 days by default; organizations can set up to 90 days for public repositories and up to 400 days for private ones (source: GitHub Docs, retention for Actions artifacts and logs). If CI results matter to a buyer, and they often do because a pass or fail is a clean outcome label, start saving run results now. For pull requests and reviews, plan an export through the API or the migration tools while the organization is active.
GitLab project export with history
GitLab's file-based project export, described in its docs page on migrating GitLab data by using file exports, includes the project and wiki repositories, issues with comments, merge requests with comments, commit comments, and archived CI/CD pipelines. It does not include CI/CD job traces and artifacts or child pipeline history, and GitLab says the list varies by version (source: GitLab Docs, migrate GitLab data by using file exports).
In practice that means a GitLab export keeps the review record (merge requests and their discussion) but not the job logs, so a buyer gets pipeline status without the detail of why a job failed. If job logs matter, save them separately while the instance is running. Self-managed GitLab keeps everything on your own servers, which makes the export a copy job rather than a race against a retention window.
Export Jira data: Atlassian terms, backups and retention
What the Atlassian Customer Agreement says
- Ownership. Section 13 says the customer owns all intellectual property and other rights in customer data and customer materials (source: Atlassian Customer Agreement (effective Oct 1, 2026)).
- Retrieval and deletion. Section 4.4 points to Atlassian's documentation for how to retrieve customer data from cloud products, and Section 12.4 says Atlassian deletes customer data after expiration or termination in line with that documentation (source: Atlassian Customer Agreement (effective Oct 1, 2026)).
- AI features. Section 4.6 places Atlassian's AI offerings under separate AI terms, which we did not review for this page (source: Atlassian Customer Agreement (effective Oct 1, 2026)).
Jira Cloud backup and export
Atlassian's page on exporting data from Jira Cloud says a site backup, made from Settings, then System, then Backup manager, can include work items and field content (system and custom fields), boards and sprints, users and groups, comments, media files and space configuration. If you include attachments, avatars or logos, you need to allow 48 hours between backups. Automation flows, third-party apps and app data, Opsgenie-powered features and Assets are not exported in the backup (source: Atlassian Support, export data from Jira Cloud).
Atlassian says app data is kept for 15 days after deactivation on free subscriptions and 60 days on paid ones, and that at the end of that period it is permanently deleted and cannot be recovered (source: Atlassian Support, reactivate a subscription). Confluence holds many of the records buyers want (design docs, RFCs, runbooks, postmortems), so back it up on the same schedule as Jira.
Linear: export all issues, and what the terms say
- Ownership. Section 10.2 of Linear's terms says the customer owns all right, title and interest in its user submissions, which Linear may use for the sole purpose of providing the service (source: Linear Terms of Service (effective Jun 9, 2026)).
- AI features. Section 10.5 says use of Linear's AI features is governed by its AI Services Addendum, which controls in a conflict (source: Linear Terms of Service (effective Jun 9, 2026)).
- Deletion. Section 5.3 says that within 30 days of termination for cause, on the customer's request after termination, or if the customer deletes its workspace, Linear will delete the customer's user information and submissions, unless the customer asks for earlier deletion. Free accounts unused for a year or more may be deleted (source: Linear Terms of Service (effective Jun 9, 2026)).
Linear's exporting data page lists CSV exports of workspace issue data (by admins, or owners on Enterprise), member lists, issues from any view, projects and initiatives, and customer requests. Members can export up to 250 issues at a time and admins up to 2,000, workspace export links expire after 12 hours, and CSV exports do not include attachment files (source: Linear Docs, exporting data). The CSV field lists on that page do not include comment threads, so if discussion on issues matters to a buyer, plan another route for it while the workspace is active.
Export Slack data, private channels included, and the API terms
| What you can export | Free and Pro | Business+ | Enterprise |
|---|---|---|---|
| Public channels (messages and file links) | Yes | Yes | Yes |
| Private channels | No | Yes, by application | Yes |
| Direct messages | No | Yes, by application | Yes |
| Scheduled recurring exports | No | Yes | No |
| Export by conversation type or member | No | No | Yes |
Slack's API terms on AI training
Slack's API Terms of Service, effective October 10, 2025, say that providers of apps used outside their own organization may not “use API Data to train a large language model” or bulk export Slack message and file data unless an additional agreement allows it. The Data Access API terms add that third-party providers may not create persistent copies or archives of other organizations' API data (source: Slack API Terms of Service (effective Oct 10, 2025)). A buyer's connector is an app from outside your organization, so a Slack license usually starts from the owner's own export.
Slack free plan history and downgrades
The free plan gives access to 90 days of message and file history, and Slack says messages and files on the free plan are stored for only one year, with older content permanently deleted (source: Slack Help Center, Pro plan pricing and Fair Billing Policy updates). Slack says deleted messages and files cannot be restored, and that canvases become read only and subject to the 90-day limit after a downgrade (source: Slack Help Center, message and file history on free workspaces). If a paid plan might lapse, as it often does during a wind-down, export private channels and DMs while you still have Business+ or Enterprise. Our guide to shutting down a startup and selling its records covers that sequence.
Export checklist and deadlines after you cancel
| System | What to export | What the standard export misses | After cancellation |
|---|---|---|---|
| GitHub | Mirror clone of every repo; PRs, reviews, issues via migration tools or API | Actions run history, status checks, comment edit history | Actions logs 90 days by default while active |
| GitLab | Project export per project | Job traces and artifacts, child pipeline history | Self-managed stays on your servers |
| Jira Cloud | Site backup with attachments | Automation flows, third-party app data | 15 days free, 60 days paid, then deleted |
| Confluence | Site or space backup | Check app data the same way as Jira | Same Atlassian retention |
| Linear | Workspace issue CSV plus projects; plan a route for comments | Attachments, comment threads in CSV | Deletion within 30 days in the cases in Section 5.3 |
| Slack | Full export on Business+ or Enterprise | Private channels and DMs on Free and Pro | Free plan: 90 days visible, one year stored |
Once the exports exist, the scope decisions are the same as for any software company: customer content, secrets, HR records and third-party code stay out, and people are replaced with consistent tokens. Our guides on selling source code to AI companies and licensing software company data cover those steps. To have the records reviewed and taken to several buyers, see how Regulr gets a company competing offers for its records, or check the factors in our data value checker.
How Regulr is paid, and how we write these pages
Regulr is a sell-side advisor. When a business we introduce signs a license, we may be paid a fee by the buyer or through a program's referral terms. We list every program we can verify whether or not it pays us, order is never set by fees, and every figure is the company's own advertised term with a link to where it says so. Regulr has not closed a data deal yet, so nothing here describes our results.
Frequently asked questions
Does GitHub train on private repos?
GitHub's terms say it treats private repository contents as confidential and will not otherwise use them to develop or improve the service. The exception is content you send to its AI features, such as Copilot, as input: on individual accounts GitHub may use it for training unless you opt out in settings. Organizations under a GitHub Customer Agreement or volume licensing agreement are governed by that agreement instead.
How do I export all my Slack data, including private channels?
On Free and Pro plans, Workspace Owners and Admins can export public channels only. Private channels and DMs need Business+ or Enterprise: on Business+, Workspace Owners apply for the expanded export and the Primary Owner approves it. If you plan to export private history, do it before any downgrade, because a free workspace shows only 90 days of history.
Can a buyer's Slack app pull our messages for AI training?
Slack's API terms say providers of apps used outside their own organization may not use API data to train a large language model or bulk export message and file data without an additional agreement. That is why licenses built on Slack history usually start from the workspace owner's own export rather than a buyer's connector. Your export is your company's copy of its own records.
How do I export Jira data with full history?
Jira Cloud's Backup manager, under Settings, then System, creates a site backup that includes work items and field content, comments, boards and sprints, and media files if you choose. If you include attachments, Atlassian says to allow 48 hours between backups. Automation flows and third-party app data are not included, so export those separately if you need them.
Can I export all issues from Linear?
Workspace admins can export workspace issue data as a CSV, and members can export issues from any view. Linear's docs set limits of 250 issues per export for members and 2,000 for admins on view exports, say workspace export links expire after 12 hours, and say CSV exports do not include attachment files. The field lists in Linear's docs do not include comments, so plan for that if comment threads matter to a buyer.
Does a GitHub repository export include pull requests and CI history?
A plain git clone with the mirror option copies the code and every branch and commit, but not pull requests, reviews or issues. GitHub's migration tools carry pull requests, issues, commit comments and Actions workflow files, but not Actions workflow run history, commit status checks or the edit history of comments. Actions logs are kept 90 days by default, so older CI results may already be gone.
How long do Atlassian, Linear and Slack keep data after we cancel?
Atlassian says app data is kept 15 days for free subscriptions and 60 days for paid ones after deactivation, then permanently deleted. Linear's terms provide for deleting customer data within 30 days in several termination cases, or sooner on request. Slack's free plan shows 90 days of history and deletes content older than a year. Export while the subscription is active.
Sources
GitHub, GitLab, Atlassian, Linear and Slack terms and documentation read on their own pages on October 18, 2026. Program terms checked October 15, 2026; our buyer directory lists the programs and links to their published terms.
- GitHub Terms of Service (effective Apr 27, 2026)
- GitHub Docs, about migrations between GitHub products
- GitHub Docs, retention for Actions artifacts and logs
- GitLab Docs, migrate GitLab data by using file exports
- Atlassian Customer Agreement (effective Oct 1, 2026)
- Atlassian Support, export data from Jira Cloud
- Atlassian Support, reactivate a subscription
- Linear Terms of Service (effective Jun 9, 2026)
- Linear Docs, exporting data
- Slack Help Center, export your workspace data
- Slack API Terms of Service (effective Oct 10, 2025)
- Slack Help Center, message and file history on free workspaces
- Slack Help Center, Pro plan pricing and Fair Billing Policy updates
- program terms, Regulr buyer directory
Brian Boesen · Founder, Regulr AI
Brian runs Regulr AI, which works for established businesses that want to license de-identified records to AI companies. He checks each program's published terms against its own pages and updates these guides when they change. Last updated October 18, 2026.
This page is general information, not legal advice, and Regulr is not a law firm. Rules vary by state, by profession and by contract, and they apply to your facts, not to a category. You approve exactly what is in scope before anything is shared.
Keep reading