{
 "public_id": "AIR-2025-0049",
 "incident_id": "2025-06-skynet-malware-prompt-injection",
 "date_disclosed": "2025-06-25",
 "date_occurred": "2025-06-01",
 "title": "'Skynet' malware prototype embedding a prompt injection to evade LLM malware analysis tools",
 "class": "in_the_wild",
 "attack_surface": "other",
 "initial_vector": "malware_embedded_prompt_injection",
 "agent_capability_abused": [
  "automated_code_analysis"
 ],
 "target_product": "LLM-based malware detection and reverse-engineering tools",
 "target_vendor": "various",
 "autonomy_level": "semi_autonomous",
 "impact_realized": false,
 "impact_type": [
  "detection_evasion"
 ],
 "blast_radius": "AI-assisted malware triage pipelines",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI01"
  ],
  "mitre_atlas": [
   "AML.CS0043"
  ],
  "labels": [
   "defense_evasion",
   "prompt_injection_against_defender",
   "first_of_kind"
  ]
 },
 "defense_present": null,
 "defense_bypassed": false,
 "fix_shipped": null,
 "primary_sources": [
  "https://research.checkpoint.com/2025/ai-evasion-prompt-injection/",
  "https://raw.githubusercontent.com/mitre-atlas/atlas-data/main/dist/v6/ATLAS-2026.07.yaml"
 ],
 "verified_quote": "In this write-up we present a malware sample found in the wild that boasts a novel and unusual evasion mechanism — an attempted prompt injection (”Ignore all previous instructions…”) aimed to manipulate AI models processing the sample.",
 "confidence": "high",
 "notes": "Check Point Research. The injection did NOT work against the models tested; impact_realized is false. Significant as the first observed in-the-wild attempt to attack defensive AI agents.",
 "risk_category": "security_fraud",
 "risk_secondary": [
  "product_misuse"
 ],
 "risk_subcategory": "Prompt Injection",
 "risk_sub_secondary": [],
 "ai_entity": "generative",
 "agency": "ai_as_target",
 "agency_debatable": true,
 "agency_reason": "the injection targets the defender's LLM analysis tool and did not work",
 "vector_family": "other",
 "vector_detail": "malware_embedded_prompt_injection",
 "guardrail_kind": [
  "unknown"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [],
  "atlas": [
   "AML.CS0043"
  ],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI01",
 "agent_framework": "unknown",
 "model_provider": "unknown",
 "model_name": "",
 "detection_mechanism": "threat_intel",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "policy_drift",
 "eai_domain": "governance",
 "eai_secondary": [],
 "eai_reason": "impact rule: policy_drift",
 "asi_curated": [
  "ASI01"
 ],
 "asi_scope": "in_scope",
 "asi_provenance": "curated",
 "asi_reason": "curated: ASI01",
 "submitted_by": "EAI Internal"
}