{
 "id": "AIR-2023-0011",
 "slug": "kaspersky-antimalware-ml-model-evaded-via-feature-knowledge",
 "title": "Kaspersky Antimalware ML Model Evaded via Feature Knowledge",
 "summary": "Researchers demonstrated that feature knowledge is sufficient to conduct adversarial attacks against ML-based malware detectors. The technique successfully evaded detection for most modified malware files without white-box access.",
 "type": "attack_on_ai",
 "categories": [
  "adversarial_evasion"
 ],
 "severity": 1,
 "targets": [
  "enterprise_deployment"
 ],
 "products": [
  "Kaspersky Antimalware ML Model"
 ],
 "incident_date": "2023-03-31",
 "incident_date_precision": "day",
 "first_reported": "2023-03-31",
 "sources": [
  {
   "url": "https://github.com/avidml/avid-db/blob/main/vulnerabilities/2023/AVID-2023-V014.json",
   "publisher": "AVID",
   "title": "AVID-2023-V014: Cloud storage and computations have become popular platforms for deploying ML malware detectors.",
   "published_at": "2023-03-31"
  },
  {
   "url": "https://atlas.mitre.org/studies/AML.CS0014",
   "publisher": "MITRE ATLAS",
   "title": "Confusing Antimalware Neural Networks",
   "published_at": "2023-03-31"
  },
  {
   "url": "https://securelist.com/how-to-confuse-antimalware-neural-networks-adversarial-attacks-and-protection/102949",
   "publisher": "Kaspersky Securelist",
   "title": "Article, \"How to confuse antimalware neural networks. Adversarial attacks and protection\"",
   "published_at": "2023-03-31"
  }
 ],
 "source_count": 3,
 "confidence": 0.9,
 "status": "published",
 "origin": "seed_avid",
 "tags": [
  "adversarial_evasion",
  "malware_detection",
  "research"
 ],
 "created_at": "2026-10-18T06:56:23.100Z",
 "updated_at": "2026-10-18T06:56:23.100Z"
}