Atlassian just switched on AI data collection for every customer account, free and paid. Your Confluence pages, Jira tickets, and other workspace data are now fair game for training their models. You're opted in automatically. The only way out is finding the opt-out setting, which multiple users say either doesn't show up in their instances or is buried so deep it might as well be hidden.
This isn't landing well. Atlassian's products have been shaky lately, with users reporting P0 bugs across Jira and Bitbucket. Community forums are swimming in thousands of bug reports and workarounds. When your software is already frustrating people, quietly turning their data into AI training fodder is a tough sell. One commenter who tried Atlassian's AI offerings during a free trial described them as non-functional, and said canceling the trial was its own headache.
The timing has people talking. Rumors about acquisition talks between Atlassian and Anthropic have been circulating. If that's real, this data collection could give Anthropic access to clean, structured business data showing how teams actually work. That's exactly what AI companies pay premiums for. Whether the acquisition rumor holds water or not, the data grab is happening now.
Atlassian is going against the current. Microsoft's terms explicitly say Azure OpenAI Service customer data won't train foundation models without consent. Google Duet AI requires opt-in. Salesforce built an entire "Einstein Trust Layer" around data privacy. Zoom tried the opt-out approach in 2023, got hammered, and reversed course fast. Everyone else is moving toward explicit consent. Atlassian just walked the other way.