“AI Data” Is on Atlassian's Own List of What Can't Be Pinned to a Region
Atlassian's own support documentation publishes an in-scope / out-of-scope table for data residency, product by product. “AI data” is in the out-of-scope column — for every app, on every plan that offers residency at all. Here's exactly what that means, and what it doesn't.
What's In Scope, and What Isn't
Atlassian's Understand data residency page opens with a clear promise: “Data residency gives you control over where your in-scope app data for Jira, Jira Service Management, Jira Product Discovery, Confluence, and Loom is hosted.” The page then publishes an in-scope / out-of-scope table per product — and the out-of-scope column is where the useful reading is.
| App | Data that can't be pinned to a region |
|---|---|
| All Atlassian apps | App analytics, AI data, Cached content, Transient data, User account information, Third-party integration data |
| Jira | AI data, Connected DevOps data (commits, branches, pull requests, builds, deployments, feature flags, remote links) |
| Confluence | AI data |
“AI data” appears in all three rows above — under All Atlassian apps, and again, separately, under both Jira and Confluence. This isn't a scope note that was missed once; it's repeated. The page also states residency “can't be set at the individual project, client, or user level within the app,” and quantifies the cost of moving regions: downtime that “may last up to 24 hours, depending on data size,” with search “temporarily disabled during re-indexing, which can take up to 3 days.”
Two Facts That Are Only Interesting Together
On its own, an out-of-scope table entry is a documentation detail. It stops being one next to a second fact: Atlassian began using Jira and Confluence data to train AI by default on August 17, 2026. The customer who bought data residency to keep Confluence in Germany did not buy residency for the AI data that same instance now contributes by default. Two Atlassian pages, read side by side, say that plainly — neither says it about the other.
Be precise about what this claim is and isn't. “Out of scope for residency” means Atlassian does not commit to pinning that data to a region. It is not a statement about which country processes it — Atlassian's page names no destination, and none should be inferred. The gap is the absence of a commitment, not evidence of a specific one.
What Residency Actually Does Cover
Residency is a real control, and it covers the part of your instance most buyers actually care about: the Confluence pages and Jira work items themselves — the in-scope app data the support page describes up front. A compliance buyer who has already read this far will check that claim and discount the whole page if it isn't here. The argument in this page is about the gap at the edge — the AI layer — not a claim that residency does nothing.
What an Admin Hasn't Priced
Two operational facts sit in the same support page, easy to skip past on the way to the scope table:
Residency is per app, not per project
Atlassian's page states residency “can't be set at the individual project, client, or user level within the app.” One region setting applies to the whole Jira or Confluence instance — there's no per-project carve-out for the one team that needs different handling.
Moving regions isn't free
A region change costs downtime that “may last up to 24 hours, depending on data size,” and search is “temporarily disabled during re-indexing, which can take up to 3 days.” Getting the region wrong the first time isn't a settings tweak — it's a scheduled outage.
Where Data Residency Came From
Data residency wasn't always available to every paying customer — it was Enterprise-gated until customers voted it open on Atlassian's own public tracker. The request, filed 9 November 2020, is now closed as Fixed with 303 votes and 200 watchers; Atlassian confirmed on the ticket that “as of late January 2022 availability of Data Residency has been rolled out to all Atlassian paid customers for the products in scope.” The AI-data gap covered above is a limit on what a customer-won feature actually reaches — not a feature Atlassian never intended to ship broadly.
This is documentation Atlassian revises without announcement — the out-of-scope table above is exactly the kind of thing that gets edited quietly. Re-verify directly against the support page before citing it in a compliance review.
Self-Managed Doesn't Have an Out-of-Scope Column
When your data never leaves your own infrastructure, there's no separate AI-data carve-out to check for — there's just where you put it. See what that looks like against your actual footprint.
Size My InstanceRelated Reading
- How to opt out of Atlassian training AI on your Jira and Confluence data →
- The Data Center end-of-life countdown, and what it means depending on your status →
- The migration gotchas that lock teams into Jira — and a 2-minute lock-in audit →
- What Jira, Confluence, and Data Center actually cost a 100-person team →
Sources
- Understand data residency — Atlassian Support
- CLOUD-11100: Provide EU based hosting for Atlassian Cloud products — Atlassian public issue tracker
Compiled August 2026 from Atlassian's own support documentation and public issue tracker. Atlassian revises this documentation without announcement — re-verify quarterly, and before citing it in a compliance review, at support.atlassian.com.