Find notable cyber news and cases, enriched with sources, timelines, and signals.

ServiceNow Now Assist second-order prompt injection via agent discovery

Technical Analysis
First reported
Last updated
Happening score
H score 16
1 unique sources, 1 articles

Summary

Hide ▲

AppOmni showed that ServiceNow Now Assist can be abused through second-order prompt injection, letting attackers drive unauthorized actions across agent teams and expose sensitive corporate data. The risk matters because the behavior can unfold under the victim user's privileges and may still succeed even when built-in prompt-injection protections are enabled.

Related Happenings

ServiceNow AI Platform unauthenticated impersonation flaw (CVE-2025-12420)

Vulnerability
First: 13.01.2026 13:47 Last: 13.01.2026 13:47 Sources 1

About this happening: **CVE-2025-12420** exposes **ServiceNow AI Platform** deployments to **unauthenticated impersonation** and **arbitrary actions**, creating a high-severity account-takeover risk. T...

ServiceNow AI Platform patch release for CVE-2025-12420

Security Patch Release
First: 13.01.2026 13:47 Last: 13.01.2026 13:47 Sources 1

About this happening: **ServiceNow** released a **security update** for **CVE-2025-12420**, a **critical** flaw in its **ServiceNow AI Platform** that could let an **unauthenticated user** impersonate...

PromptSteal and PromptFlux AI-enabled malware activity

Malware Activity
First: 06.11.2025 11:45 Last: 06.11.2025 11:45 Sources 1

About this happening: **PromptSteal** and **PromptFlux** now show how malware can use **LLMs during execution** to generate malicious code on demand, raising the risk of more adaptive evasion and theft...

Timeline

  1. 19.11.2025 11:59 2 articles · 6mo ago

    ServiceNow Now Assist default settings enable second-order prompt injection

    Technical Analysis Update

    AppOmni described how default configurations in ServiceNow Now Assist can be abused for second-order prompt injection through agent-to-agent discovery and collaboration, allowing a malicious prompt embedded in accessible content to redirect a benign agent into unauthorized actions such as copying or exfiltrating sensitive corporate data, modifying records, sending emails, or escalating privileges. The behavior can unfold under the privileges of the user who started the interaction, and ServiceNow later said the behavior is intended and updated its documentation for clarity; recommended defenses include supervised execution mode for privileged agents, disabling sn_aia.enable_usecase_tool_execution_mode_override, segmenting agent duties by team, and monitoring AI agents for suspicious behavior.

    Show sources