Atlassian's New Data Contribution Policy: What It Means for Your Organisation

Ed Rouse

20 April, 2026

Key Points

From 17 August 2026, Atlassian will use metadata and in-app data from your Jira, Confluence, and other cloud products to train its AI models
What you can opt out of depends on your subscription tier. Only Enterprise customers can opt out of metadata collection
Some organisations are automatically excluded from data collection entirely (government cloud, HIPAA, certain financial services)
Settings are available to review now in Atlassian Administration under Security > Data contribution
You have until 17 August to review and adjust your defaults

Summary

If you’re running Jira, Confluence, or any other Atlassian cloud product, there’s a policy update coming that’s worth a few minutes of your time. It’s not a crisis, but it does involve some settings that will be switched on by default from 17 August 2026, and the defaults vary depending on your subscription tier.

From that date, Atlassian will start using customer data to improve its AI capabilities. That covers metadata from your Jira projects and Confluence spaces, and depending on your plan, potentially the actual content your team creates day to day.

Here’s the plain-English version of what’s changing, what you can control, and what it might mean for your organisation.

What Atlassian Is Collecting

There are two types of data involved.

Metadata is the structural and behavioural information around your content. Think readability scores for Confluence pages, task classifications, semantic similarity scores between documents, story points on Jira issues, sprint end dates, and JSM SLA values. Atlassian describes this as de-identified and aggregated before use, with direct personal identifiers such as names and email addresses removed. You can see the full list of what counts as metadata in Atlassian’s own documentation.

In-app data is the actual content your team creates: Confluence page titles and body content, Jira issue titles, descriptions and comments, custom emoji names, custom workflow names, and custom status labels. This is your business context: project names, technical decisions, process documentation, and internal discussions.

Both types will be used to improve Atlassian’s AI capabilities across the platform, including Rovo, and will be retained for up to seven years.

What You Can Control, and What You Cannot

Your ability to opt out depends on which plan you’re on. The table below summarises the defaults and what can be changed. Full details are in Atlassian’s data contribution settings documentation.

Free plan
- Metadata always contributed, no opt-out available
- In-app data on by default, can be turned off
Standard plan
- Metadata always contributed, no opt-out available
- In-app data on by default, can be turned off
Premium plan
- Metadata always contributed, no opt-out available
- In-app data off by default, can be turned on
Enterprise plan
- Metadata on by default, can be turned off
- In-app data off by default, can be turned on

Only Enterprise customers can opt out of metadata collection. All other tiers, including Premium, contribute metadata automatically. Full control over both data types requires Enterprise licensing or specific technical configurations such as customer-managed encryption keys (CMK or BYOK).

Automatic Exclusions

Some organisations are excluded from data collection entirely, regardless of plan tier. You do not need to take any action if your organisation falls into one of these categories:

Atlassian Government Cloud customers
Atlassian Isolated Cloud customers
Organisations with HIPAA compliance requirements
Certain government and financial services customers

If you are unsure whether your organisation qualifies for an automatic exclusion, check your settings in Atlassian Administration or contact your account team.

How Your Plan Tier Is Determined

One thing worth knowing: your data contribution settings are not determined per-product. They apply at the organisation level, based on the highest active plan across your entire Atlassian organisation, including any trials.

A few practical examples:

Confluence Free + Jira Standard: Standard tier rules apply
Confluence Standard + Jira Premium trial: Premium tier rules apply
Confluence Free + Jira Standard + Jira Product Discovery Premium: Premium tier rules apply
Confluence Free + Jira Enterprise: Enterprise tier rules apply and metadata is controllable

A single Enterprise licence in your organisation changes the settings for everything in that organisation. If you are on Premium across the board, you have more in-app data control than Standard customers, but metadata contribution still cannot be turned off.

A Note on the Contract Side

The new data contribution policy takes effect alongside an updated Atlassian Customer Agreement, also effective 17 August 2026. Atlassian has updated its terms to explicitly permit this use of customer data, so this is a coordinated change rather than something slipped through the back door. The updated AI Terms are also effective from the same date.

A few things are worth understanding if you manage enterprise contracts or advise customers on their Atlassian setup.

Data ownership. The customer agreement states clearly that customers own their Customer Data. Atlassian’s position is that de-identified, aggregated metadata falls outside that definition in a meaningful sense, which is why it can be collected without an opt-out for lower tiers. That framing is reasonable but worth being aware of.

Mid-term changes. The agreement allows Atlassian to push modifications into a current subscription term when they reflect updates to product functionality or new features. Customers who object can terminate the affected subscription term and receive a pro-rated refund. This is standard SaaS contract practice, but it does mean there is no mechanism to stay on existing terms without accepting the change.

Your own obligations. This one is easy to overlook. The agreement places an obligation on customers to have made all necessary disclosures and obtained all required consents for Atlassian to use Customer Data. If your employees are creating content in Confluence and Jira and that content feeds AI training, you may have notification obligations under your own applicable privacy legislation, including the New Zealand Privacy Act, the Australian Privacy Act, GDPR, or others. Those obligations sit with you, not Atlassian, so it is worth a conversation with your legal or privacy team before August if this is a concern.

What to Do Before August 17

Check your org’s settings. Go to Atlassian Administration, then Security, then Data contribution. Settings have been rolling out since 16 April, so they should be available to review now.
Know your tier. Understand which plan is the highest active plan in your organisation, including any trials, because that determines your defaults.
Exclude sensitive spaces or apps if needed. If in-app data contribution is on for your plan tier, you can selectively exclude specific Confluence spaces, Jira projects, or Teamwork Graph connectors. This gives you granular control even when you cannot turn off the setting entirely.
Review your privacy notices if relevant. If your organisation has internal policies or employee disclosures covering how business tools handle data, consider whether an update is appropriate.
Consider whether Enterprise changes the calculation. For organisations where data governance is a serious consideration, including government, financial services, legal, and healthcare, the metadata opt-out available at Enterprise tier may now be a more compelling reason to consider uplift than AI features alone.
Watch Atlassian’s webinar recording. Atlassian ran a customer briefing on 28 April 2026 covering the changes in detail. A recording is expected to be made available via Atlassian’s data contribution page.

The Bigger Context

Atlassian is not alone in making this kind of move. Microsoft, Salesforce, and Google are all heading in the same direction, using customer data at scale to improve AI capabilities. The trade-off Atlassian is making explicit here is that better training data produces better AI, and that should return value to customers over time.

The practical reality for most organisations is that the defaults are set to contribute, and you will need to take active steps to change them if that matters to you. August 17 is when the policy takes effect, not a deadline for decision-making. There is time to review your settings and make deliberate choices about what your organisation contributes.