Prompted Rovo actions can exfiltrate Jira and Confluence data to attackers

Intelligence Summary

Atlassian’s Rovo assistant can be manipulated to collect Jira or Confluence content from a signed-in user. The retrieved information can then be sent to an attacker-controlled external server. Two separate security firms observed the behavior using different prompt or instruction routes. One exploitation path has been confirmed as closed, but the full fix status is still uncertain. Users should limit sensitive content exposure in connected Atlassian spaces and monitor for unusual data access and outbound activity.

Recommended Action

Review affected assets, schedule urgent remediation, and monitor related indicators.

Topics