The Key Is in "By Default"
The focus of the controversy isn't whether to collect, but how consent is obtained: the option is checked by default, buried deep in the admin backend, with a vaguely worded announcement, and by the time users noticed, data had already been flowing for a while. What's stored in Jira and Confluence? Project plans, incident postmortems, customer information, internal debates—all a company's most sensitive operational data. Using this kind of data to train models is hard for compliance departments to nod at, even if the vendor promises de-identification and aggregation—let alone that many customers also carry industry regulatory requirements.
Data Anxiety in the SaaS Era
Atlassian isn't the first—Adobe, Zoom, and Slack have all stepped on the same PR landmine, and the script is identical: quietly change the terms, get caught, apologize, and move the toggle somewhere more visible. The frequency is high enough to show this is a systemic industry impulse—every SaaS company sits on unique industry data, and in the AI era this data suddenly became a gold mine, too tempting to resist. The pragmatic response is just three things: go check the relevant settings in your own Atlassian backend now, write data-training terms into the review checklist for procurement contracts, and default to assuming every SaaS vendor has the same impulse and inspect periodically.
via: Hacker News