ServiceNow Now Assist Exploit Enables AI Prompt Injection

Security researchers at AppOmni have discovered a second-order prompt injection vulnerability in ServiceNow’s Now Assist platform. The ServiceNow Now Assist exploit leverages default settings that automatically group AI agents into discoverable teams, allowing attackers to seed hidden instructions in data fields. When one agent processes this malicious prompt, it can recruit other AI agents in a coordinated chain reaction, leading to unauthorized data theft, record modification, and privilege escalation. Because the flaw stems from standard configurations rather than a coding bug, many organizations may be unaware their ServiceNow Now Assist deployment is at risk. To mitigate this exploit, security teams should audit default AI settings, disable automatic agent discovery, enforce strict access controls, and implement continuous monitoring of AI workflows. Prompt remediation is essential to prevent attackers from exploiting these new AI-driven attack vectors across SaaS environments.