{
 "public_id": "AIR-2025-0177",
 "incident_id": "2025-11-servicenow-agent-discovery-injection",
 "date_disclosed": "2025-11-19",
 "date_occurred": null,
 "title": "ServiceNow Now Assist: second-order prompt injection recruits privileged agents via agent-to-agent discovery",
 "class": "responsible_disclosure",
 "attack_surface": "multi_agent_system",
 "initial_vector": "indirect_prompt_injection_ticket",
 "agent_capability_abused": [
  "agent_to_agent_discovery",
  "record_crud",
  "outbound_email"
 ],
 "target_product": "ServiceNow Now Assist AI Agents",
 "target_vendor": "ServiceNow",
 "autonomy_level": "semi_autonomous",
 "impact_realized": false,
 "impact_type": [
  "data_exfiltration",
  "privilege_escalation",
  "data_integrity_loss"
 ],
 "blast_radius": "ServiceNow instances running Now Assist agents with default team grouping and discoverability",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI01",
   "ASI07",
   "ASI03"
  ],
  "mitre_atlas": [],
  "labels": [
   "second_order_injection",
   "agent_to_agent",
   "confused_deputy",
   "privilege_escalation",
   "insecure_defaults"
  ]
 },
 "defense_present": "ServiceNow prompt-injection protection was enabled during the demonstration",
 "defense_bypassed": true,
 "fix_shipped": "none; ServiceNow confirmed the behaviors were intended and updated platform documentation instead",
 "primary_sources": [
  "https://appomni.com/ao-labs/ai-agent-to-agent-discovery-prompt-injection/"
 ],
 "verified_quote": "I instructed a seemingly benign Now Assist agent to recruit more powerful agents in fulfilling a malicious and unintended task.",
 "confidence": "high",
 "notes": "AppOmni Labs (Aaron Costello). Published 2025-11-19. A low-privileged user who can only create tickets plants hidden instructions; a benign agent later reads them and recruits more privileged agents that act with the initiating admin's authority. ServiceNow 'confirmed these behaviors were intended'; this is a configuration/default-posture failure, not a patched bug, which makes it a standing risk rather than a closed one. Root causes cited: insecure LLM selection and default team-based agent grouping.",
 "risk_category": "security_fraud",
 "risk_secondary": [
  "privacy"
 ],
 "risk_subcategory": "Prompt Injection",
 "risk_sub_secondary": [
  "Unauthorized Access & Account Takeover",
  "Multi-agent contagion"
 ],
 "ai_entity": "generative",
 "agency": "agent_acted",
 "agency_debatable": false,
 "agency_reason": "the Now Assist agent recruited privileged agents and emailed data out",
 "vector_family": "indirect_prompt_injection",
 "vector_detail": "indirect_prompt_injection_ticket",
 "guardrail_kind": [
  "unknown"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [],
  "atlas": [],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI01",
 "agent_framework": "unknown",
 "model_provider": "unknown",
 "model_name": "",
 "detection_mechanism": "unknown",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "privilege_escalation",
 "eai_domain": "access_control_and_permissions",
 "eai_secondary": [
  "data_exfiltration_channels",
  "confused_deputy"
 ],
 "eai_reason": "impact rule: privilege_escalation",
 "asi_curated": [
  "ASI01"
 ],
 "asi_scope": "in_scope",
 "asi_provenance": "curated_plus_rule",
 "asi_reason": "curated: ASI01 | added ASI07 (vector=inter_agent), ASI03 (signal=identity)",
 "submitted_by": "EAI Internal"
}