🟠 High | Source: The Hacker News
Atlassian’s AI assistant Rovo can be manipulated via prompt injection attacks to exfiltrate Jira and Confluence data accessible to a signed-in user, sending it to an attacker-controlled server. Two independent security firms discovered separate attack routes; at least one remains unpatched. This is a significant risk for organisations using Rovo across sensitive project management and documentation platforms.
Security Architect’s Take: Review and restrict Rovo’s access to sensitive Jira projects and Confluence spaces immediately, and audit what external content Rovo is permitted to read. Consider disabling Rovo in high-sensitivity environments until Atlassian confirms all identified attack vectors are fully remediated.
Original advisory: Atlassian Rovo Can Be Tricked Into Sending Jira and Confluence Data to Attackers