Atlassian's AI tool Rovo is susceptible to zero-click data exfiltration via indirect prompt injection. This vulnerability allows the system to bypass organization-level web search controls and exfiltrate data without requiring user interaction.