Atlassian Rovo Can Be Tricked Into Sending Jira and Confluence Data to Attackers

Atlassian's Rovo assistant can be deceived into collecting and sending Jira or Confluence data to attackers. This can occur when attacker-controlled instructions are executed, allowing the assistant to gather data accessible to a signed-in user and transmit it to an outside server. Two separate security firms discovered this vulnerability through different methods. However, only one of these methods has been confirmed as closed. One of the security firms, PromptArmor, was able to hide the instructions in content that Rovo reads, using an uploaded file. The vulnerability was discovered independently by the two firms, highlighting a potential security risk.