Atlassian will mine most customers’ data for AI — unless you pay up

What’s changing

It has been reported that, starting 17 August 2026, Atlassian will begin collecting customer metadata by default to train its AI models — unless customers are on the top enterprise tier or legally exempt. The company’s footprint is large: roughly 300,000 customers use Jira, Confluence and other cloud products, and Atlassian says the metadata it will gather includes readability and complexity scores, task classifications, semantic-similarity measures between pages, and numeric fields such as story points, sprint end dates and SLAs. The claimed aim? Better search, smarter summaries, and more useful agentic workflows. Nice in theory. Expensive in practice.

Who pays, who gets protected

Allegedly, lower-paying customers have no choice about metadata collection. “If an Atlassian customer's highest active plan is Free, Standard, or Premium, metadata contribution is always on, and they are not able to opt out,” Arseny Tseytlin, head of product communications at Atlassian, told The Register. In‑app content data (titles, descriptions, comments, custom names and the like) behaves differently: Free and Standard tiers have it on by default but can opt out; Premium and Enterprise have it off by default. Some customers are entirely excluded from contribution — those using customer‑managed keys (BYOK), Atlassian Government Cloud, Isolated Cloud, HIPAA-bound customers, and certain government and financial services users.

Retention and rollback

Atlassian says collected metadata will be de‑identified, aggregated, and retained for up to seven years — long enough, it argues, to spot trends over time. If you opt out after 17 August, or delete apps, Atlassian alleges it will remove corresponding in‑app data from its datasets within 30 days. The policy won’t be enforced until the August date; if you terminate now, you won’t be covered by the new settings. Is privacy becoming a luxury feature for those who can afford enterprise pricing? That’s the uncomfortable question here, and one that will keep small teams awake as the AI era reshapes vendor economics.

Sources: The Register