Your work data is now a training resource.
Starting August 17, 2026, Atlassian will automatically collect metadata and in-app content from Jira, Confluence, and other cloud products to train its AI models — specifically its Rovo AI suite. The opt-out window is open between now and May 19, 2026, when Atlassian begins rolling out the new settings in Atlassian Administration. After August 17, if you haven’t touched those settings, your data is going in.
This is the part where I’d normally say “to be fair to Atlassian” — but I’m not sure fairness is the right frame here. What Atlassian is doing is legal, increasingly common, and still worth calling out clearly: they flipped the default switch to “yes” without asking first.
What’s Actually Being Collected
According to the verified policy details, Atlassian is collecting customer metadata and in-app content from its cloud products. That means the tickets your team writes, the Confluence docs your engineers maintain, the project structures your org has spent years building — all of it is potentially in scope. This isn’t some abstract telemetry ping. This is the operational DNA of your company.
For a lot of teams, Jira and Confluence aren’t just productivity tools. They’re where sensitive product roadmaps live. Where post-mortems get written. Where internal decisions get documented. The idea that this content feeds an AI training pipeline by default — without a single person at your company actively agreeing to it — is a real problem.
The Plan Tier Problem
Here’s where things get more uncomfortable. Free and Standard plan users are opted in by default, with no clear path to full opt-out. Some data collection reportedly cannot be opted out of at all on certain plans. If you want more control, you need to be on a higher tier — which, conveniently, costs more money.
So the structure is: pay less, give more data. Pay more, get to say no. That’s not a privacy policy. That’s a pricing strategy dressed up as one.
Smaller teams and startups — the exact users most likely to be on Free or Standard plans — are the ones with the least use here. They’re also the ones least likely to have a dedicated admin watching for policy changes buried in product update emails.
The Opt-Out Window Is Real, But Narrow
To Atlassian’s credit, they are providing an opt-out mechanism, and the rollout of settings in Atlassian Administration is happening gradually between now and May 19, 2026. That gives admins a window to act before the August 17 collection date kicks in.
But “opt-out by a specific date or we take your data” is a consent model that benefits the collector, not the user. Genuine consent is opt-in. What Atlassian is doing is the digital equivalent of signing you up for a mailing list and telling you to unsubscribe if you don’t want it.
If you’re an admin at a company using Atlassian cloud products, the action item is simple: get into Atlassian Administration before May 19, 2026, find the data contribution settings, and make an active choice. Don’t let the default decide for you.
The Bigger Pattern Worth Watching
Atlassian isn’t alone in this. Across the SaaS space, companies are quietly updating terms and flipping defaults to feed their AI ambitions. What makes the Atlassian case notable is the scale — Jira and Confluence are used by millions of teams globally — and the specificity of what’s being collected. This isn’t usage analytics. This is content.
The AI arms race has created a data hunger that’s pushing companies to extract value from existing user bases rather than build clean, consensual data pipelines. That’s a shortcut, and users are the ones subsidizing it.
Rovo is Atlassian’s bet on AI-assisted work — search, summarization, agents that can act across your tools. To build something like that, you need training data, and your Jira backlog is apparently a solid source. The business logic is clear. The ethics are murkier.
What You Should Do Right Now
- Log into Atlassian Administration and review your data contribution settings before May 19, 2026.
- Check which plan tier your organization is on — Free and Standard users have the least control.
- Flag this to your legal or compliance team if your Confluence or Jira instance holds sensitive internal content.
- Decide actively. Don’t let inaction be your answer.
Atlassian may have every right to do this. But you have every right to say no — at least while that option still exists.
🕒 Published: