Hidden text in a PDF is enough to steal sensitive data through Atlassian's AI agent Rovo

Security researchers at PromptArmor discovered that malicious instructions hidden within a PDF can hijack Atlassian's AI agent Rovo to exfiltrate sensitive data from Jira and Confluence. Because the process executes silently without requiring user authorization or leaving a record of the breach, it poses a significant risk to organizations that use the tool to index internal documents. This vulnerability highlights the security challenges associated with granting AI agents autonomous access to sensitive enterprise databases.