At DEF CON 34 on August 8, Varonis Threat Labs disclosed a critical one-click vulnerability in Atlassian’s Rovo AI assistant, named RovoBlast, which could hijack a user’s live AI session and exfiltrate sensitive enterprise data via a single crafted link. Atlassian patched the flaw before publication, and the attack required no jailbreak and no permission bypass. The vulnerability exploited a URL parameter called rovoChatPrompt that pre-fills content into Rovo’s chat window without a user-facing warning; attackers could leave the organization ID portion of the URL blank, and Atlassian would still route the request into the victim’s default organization. Once clicked, malicious instructions were seeded as trusted input. Rovo’s access across a typical enterprise deployment included Jira, Confluence, Bitbucket, Slack, Google Workspace, Microsoft 365, relational databases, uploaded files, web pages, and archived content. Data exfiltration relied on ResearchAgent, a built-in autonomous tool that pulled internal data and pushed it to an external destination in a single automated chain. Varonis demonstrated three proof-of-concept scenarios, stealing Confluence pages, Jira tickets, and SharePoint content containing personal data. The technique required no chaining of multiple requests. Varonis classifies this as parameter-to-prompt injection, the same category reported in Microsoft Copilot under the name Reprompt in January 2026. Atlassian acknowledged