- Atlassian will activate data collection for AI training in Jira and Confluence by default starting in August 2026, although it allows for its deactivation.
- Rovo expands knowledge search by integrating external repositories such as Google Drive and SharePoint while strictly respecting user permissions.
- AI governance is evolving towards a blocklist model to control which applications have access to corporate information.
- The massive deployment of AI is accelerating the end of life for Data Center products, setting the final closure for March 2029.
If you work with the Atlassian ecosystem, you've probably already noticed that artificial intelligence has gone from being a futuristic promise to the driving force behind Jira and Confluence. It's not just about adding a chat button, but a profound restructuring of how we manage corporate knowledge , where AI is now able to read, summarize, and connect data that was previously buried in digital silos.
However, this technological leap has sparked intense debate about privacy and governance. The most talked-about development is the change in data usage terms, which forces us to question whether we are simply rushing the adoption of AI or if we truly have control over what sensitive information the machine can process and who is in charge of those decisions.
The shift in AI data policy and training
Starting August 17, 2026, Atlassian will implement a significant change: default data collection in Jira and Confluence to feed its AI models will be enabled. This means that metadata capture will become mandatory for the Free, Standard, and Premium plans. While internal content such as issues, documents, and comments will also be enabled by default, organizations have the option to disable this feature, except in specific environments such as HIPAA-compliant ones where exceptions exist.
It is important to emphasize that Atlassian applies a data anonymization process , although data can be stored for up to seven years. For many experts, this appears to be a "privacy for a price" model, where regulatory compliance and data protection become assets requiring more active management by the administrator.
Atlassian Intelligence and the Rovo ecosystem
To understand where we're headed, we need to differentiate between the two layers of AI the company offers. On one hand, we have Atlassian Intelligence , a service integrated into the Premium and Enterprise licenses. This tool allows us to translate natural language into SQL or JQL, generate text drafts in the editor, summarize endless comment threads, and create automations without programming knowledge—simply by typing what we want to happen.
On the other hand, there's Rovo , a standalone product that raises the bar. Rovo doesn't just rely on Atlassian data; it uses connectors to read information from external repositories like Google Drive, SharePoint, Slack, GitHub, or a self-hosted alternative to Notion . This breaks down departmental barriers, allowing a user to find data in a Dropbox PDF without leaving their Jira workflow.
Governance: From the whitelist model to the blacklist model
One of the most disruptive changes in cloud management is the shift from an allowlist to a blocklist model . Previously, the administrator decided which applications could use AI; now, the focus is on defining which applications should NOT have access to Rovo. This reflects a change in mindset: AI is no longer an extra feature, but a critical infrastructure and a risk vector that must be strictly governed.
This new capability allows for centralized control by management, providing real-time visibility into where AI is active and preventing automated agents from accessing confidential information. The question is no longer how to adopt AI faster, but where it is dangerous to use it and who has the authority to set those limits.
Security, permits and the operation of the Agents
There's a recurring fear that AI will "learn" from our corporate secrets to help other companies. Atlassian is adamant about this: your organization's data is not used to train global models for OpenAI, Mistral, or Claude. Thanks to strict confidentiality agreements, what happens within your instance stays within it.
Furthermore, Rovo adheres to the "golden rule" of permissions: it only sees what the user is authorized to see . If you don't have access to a private Confluence space, the AI won't either. As for Rovo Agents, which can perform tasks such as creating incidents or auditing reports, their execution always depends on the permissions of the user who activates them, ensuring that there are no unauthorized privilege escalations .
Rovo's abilities and real limitations
Rovo's capabilities allow for the automation of reproducible workflows using slash commands (such as /create-jira-issueThey are ideal for standardize structured content or analyze the sentiment of thousands of support tickets. However, it's not magic. Because they rely on probabilistic models, they can generate incomplete answers or even fabricate data that seems reasonable but is actually false.
Therefore, its use is discouraged in situations requiring absolute and deterministic precision , such as in contractual or regulatory texts. Human oversight is essential; AI should be seen as an assistant that delivers a draft, not as a substitute for professional judgment. The quality of the result will always be linked to the quality of the prior documentation: if the knowledge base is chaotic, the AI will simply summarize that chaos.
The path to the cloud and the end of the data center
This entire AI deployment is the reason Atlassian has set the end-of-life date for its Data Center products for March 28, 2029. The strategy is clear: the future is cloud-first. Starting in March 2026, new organizations will be born directly in the cloud, as it is the only environment where AI can be iterated at the speed the market demands.
To facilitate this transition, the Ascend program offers incentives and support, recognizing that migration is not just a technical process, but a strategic shift . Moving to the cloud enables access to AI governance, advanced automation, and operational cost reductions that on-premises server maintenance can no longer match.
The transition to an AI-powered environment at Atlassian means accepting that data management is no longer static, but dynamic and conversational. While tools like Rovo and Atlassian Intelligence eliminate friction in information retrieval and content creation, they shift the responsibility to administrators to ensure that privacy and regulatory compliance are not sacrificed in the name of productivity. The success of this implementation will depend not on the technology itself, but on organizations' ability to document their processes and govern access to information intelligently.
Passionate writer about the world of bytes and technology in general. I love sharing my knowledge through writing, and that's what I'll do on this blog, show you all the most interesting things about gadgets, software, hardware, tech trends, and more. My goal is to help you navigate the digital world in a simple and entertaining way.