· via dev.to (home feed)
Atlassian AI data sharing defaults to on for Free and Standard plans
Since August 2026, Atlassian has used Jira and Confluence content to improve its products by default on Free and Standard plans. Admins must actively switch it off, and metadata still flows on everything below Enterprise.

Opt-out, not opt-in
Organizations running Jira and Confluence on Atlassian's Free or Standard plans have been contributing their content to product improvement since 17 August 2026, unless an administrator actively switched a single setting off. That is the picture that emerges from a hands-on walkthrough published on dev.to, which traced the defaults, the documentation and the admin experience on a live organization.
The change stems from Atlassian's updated Data Processing Addendum, effective 17 August 2026. According to dev.to, it states that Atlassian must de-identify and aggregate customer data and may use it to improve its cloud products, in line with the contribution settings.
Where the setting lives and why it is easy to miss
The control is called Data contribution, and it sits under Security in Atlassian Administration — one setting for the whole organization. Only an Organization admin can change it, it does not exist on Atlassian Government Cloud, and companies with multiple organizations must repeat the change in each.
Discovery appears to be the bigger problem. According to dev.to, a Community thread posted on 21 April warning that Atlassian had enabled customer data collection by default had drawn only 3,863 views by 16 November, and most of what ranks in search on the topic was written before the policy took effect.
In-app data versus metadata
Atlassian separates contribution into two streams, and they are treated very differently.
In-app data is the material people actually write: Confluence page titles and content, Jira work item titles, descriptions and comments, plus custom status and workflow names.
Metadata is derived rather than written, and comes in two kinds. Content attributes include a readability score for a page, how similar two pages are, classifications such as flagging a work item as sales-related, story points, and SLA values on service management requests. Common patterns cover signals spotted across many customers, including search queries, Rovo Chat prompts and configuration items. Atlassian says it discards low-frequency data that could be unique to an organization.
Per Atlassian documentation cited by dev.to, metadata feeds work such as training the search model and root cause analysis suggestions. Data from one product can improve features across the portfolio, and Rovo Chat conversations and search queries fall under the same setting.
Defaults depend on the highest active plan
On Free and Standard plans, in-app data contribution defaults to on and can be switched off. On Premium and Enterprise it defaults to off. Enterprise additionally receives a metadata switch — the only plan with one — though metadata itself still defaults to on. On Free, Standard and Premium, metadata is always contributed with no opt-out at all.
The deciding factor is the highest active plan anywhere in the organization, trials included. Confluence Free combined with Jira Standard means in-app data is on, while adding a Jira Premium trial flips it off, according to Atlassian's own examples. A single Enterprise app unlocks the metadata switch for the entire organization, and on Enterprise that switch also governs metadata from applicable Teamwork Graph connectors, which pull content from other tools into Atlassian's apps. Atlassian cautions that customers with specific compliance requirements may see different defaults, so the settings page itself is the source of truth.
Switching it off, and an audit-log gap
dev.to tested the flow on a Premium organization on 16 November. Selecting the off option does nothing until a separate Save button is pressed, and there is no confirmation dialog. The organization's audit log recorded the change in the same minute — but there is a catch. Per Atlassian's documentation, organizations without Atlassian Guard, Atlassian's paid security add-on, or an Enterprise plan do not store organization-level audit events at all. For those customers, flipping the setting leaves no trace, which matters if they later need to prove when the change was made.
A middle path exists: keep the organization opted out and include specific spaces, apps or connectors one by one, as the page's guided tour explains. Platform apps such as Rovo and Goals cannot be selected individually and simply follow the organization-level setting, and neither can content that sits outside any space.
Why it matters
The default here is opt-out, not opt-in: on the two cheapest plans, the content teams write in Jira and Confluence feeds Atlassian's improvement pipeline unless someone notices and acts, and metadata — which can encode story points, SLA figures and usage patterns — has no opt-out below Enterprise regardless. For administrators, the practical steps are to check the Data contribution page, switch off in-app data if that matches policy, and document the date themselves rather than trusting the audit log to remember it.
- #atlassian
- #jira
- #confluence
- #ai-training
- #data-privacy