As enterprise AI moves deeper into the systems that run strategy, delivery, and operations, leaders need to ask a critical question: Whose advantage is this AI actually built to serve? Atlassian’s newly announced data contribution policy puts that question front and center for every organization that uses work management platforms to plan roadmaps, manage delivery, and capture institutional knowledge.
What is Atlassian’s data contribution policy for AI training? Beginning August 17, 2026, Atlassian will begin using eligible customer metadata and in-app data to improve apps and AI experiences for all customers.
This is not a routine administrative update; it is a statement of product philosophy. Contributed customer data is now being used to improve experiences across the broader market. For enterprises managing sensitive product direction, transformation plans, or regulated delivery environments, this raises significant strategic concerns.
The blurring of competitive boundaries
Atlassian frames this approach as responsible, noting that data is de-identified and aggregated before being used across customers. They also note that for “common patterns,” they extract data that is commonly used among multiple organizations, while omitting low-frequency, unique data.
Those safeguards are important, but they do not eliminate the core issue. Even when raw content is not directly exposed, cross-customer learning shapes better recommendations, smarter templates, and more capable AI assistants. These improvements can ultimately benefit your direct competitors. According to Atlassian’s support documentation, contributed metadata can include story points, sprint end dates, readability scores, SLA values, search queries, and prompt responses. In-app data can include page titles, descriptions, and workflow names.
Enterprise work management systems do not just store tickets and pages. They hold investment priorities, process patterns, and the connective tissue between strategy and execution. This data is not administrative residue; it is a living record of how your business competes.
The default settings dilemma
Should a paying customer’s system of record become the training input for product experiences that improve outcomes for everyone else? Atlassian’s default settings make this a pressing question.
According to Atlassian’s published defaults, organizations on Free and Standard plans have metadata always contributed and in-app data on by default. Premium tier customers have metadata always contributed and in-app data off by default. Only Enterprise customers can turn metadata contribution off.
In other words, for a large portion of the market, metadata contribution is not optional. The concern here is not only privacy; it is competitive separation. Do you want your vendor’s AI roadmap to be advanced by learning from your operating model and execution signals?
The ValueOps by Broadcom standard: Customer control first
How do the AI data privacy policies of ValueOps by Broadcom and Atlassian differ? ValueOps applies a fundamentally different standard. At Broadcom, we believe enterprise AI should be grounded in customer control, governed context, and measurable business value.
AI should help your organization make better decisions and accelerate execution using your strategic and operational context. It should not require you to quietly accept cross-customer contribution as the default price of innovation.
This principle is foundational to how Broadcom positions ValueOps and Rally. Broadcom’s architecture ensures that the large language model (LLM) is not training on customer data. By default, Rally’s AI module is blocked from accessing data specific to your subscription unless that access is explicitly enabled by you. This is a materially different posture from a model centered on using contributed customer data to improve AI experiences for everyone. (Find out how Rally now features Vaia’s contextual AI, and how it can help you translate top-level strategy into a fully aligned backlog.)
Protecting your advantage
In a work management platform, AI should strengthen your business’ ability to execute strategy, not blur the lines around who benefits from your data.
If your roadmap contains your next growth move, modernization bet, or market response, the standard should be simple: Your system of record should work for you, not for everyone else. That is the standard we believe in at Broadcom, and it is why enterprises should take a fresh look at Rally and the broader ValueOps platform.
Frequently asked questions
Q: Which Atlassian tiers allow users to opt out of metadata contribution?
A: Only Enterprise customers have the ability to turn off metadata contribution. For Free, Standard, and Premium tiers, metadata contribution is always on and not optional.
Q: What specific types of data does Atlassian collect under the new policy?
A: Atlassian may collect such metadata as story points, sprint dates, and SLA values. In addition, it may also gather in-app data like page titles, workflow names, and prompt responses.
Q: How does Rally’s approach to AI data privacy differ?
A: By default, Rally’s AI module is blocked from accessing any data specific to your subscription. Access must be explicitly enabled by the user, ensuring the LLM does not train on your sensitive data.
Q: Does de-identifying data eliminate the competitive risk?
A: Not entirely. Even if raw content is de-identified, cross-customer learning allows AI models to create smarter templates and recommendations. By learning your operating model, AI can make improvements that benefit your direct competitors.
