Atlassian Shifts AI Strategy, Making Customer Data the Default Fuel

Atlassian is changing the rules on how it uses customer information. Starting August 17, 2026, the company will begin collecting metadata and content from its cloud products, including Jira and Confluence, to train its AI systems. This move represents a significant departure from previous assurances, directly enabling features like Rovo and Rovo Dev.

Administrators will see new settings appear on April 16, 2026, with a short window to adjust preferences before defaults are applied in May. The policy affects the core tools used by an estimated 300,000 organizations globally. Atlassian states it will de-identify the data, stripping out names and emails, and aggregate it at the customer level before use.

The default settings, however, vary sharply by subscription tier. Free and Standard plan users will have both metadata and in-app content collection enabled by default, with no option to disable metadata sharing. Premium plans start with in-app content collection off, but metadata sharing remains mandatory. Only Enterprise customers retain full control, with both data categories opted out by default.

The company argues this approach is necessary to build genuinely useful AI. Product communications lead Arseny Tseytlin told The Register that aggregated data allows for "more meaningful observations over longer periods of time," aiming to improve search, summarization, and workflow automation. Data may be retained for up to seven years for analysis.

Reaction from the user community has been pointed. Critics on social media and forums describe the tiered structure as a 'privacy paywall,' where smaller companies effectively subsidize AI development for larger enterprises that can afford to opt out. The mandatory metadata collection for lower-tier plans—which can include project velocity, workflow patterns, and classification data—is a particular concern, even in de-identified form.

This strategic pivot places Atlassian alongside other major SaaS providers leveraging user data for AI, testing the balance between innovation and user trust. Administrators are advised to review their organization-level settings in Atlassian Administration before the May deadline.

Source: Webpronews