EvidentAI — Audit & Compliance

MITRE ATLAS

v5.4.0
Reconnaissance 2
AML.T0000 Search for Victim's Publicly Available Research MaterialsAML.T0040 ML Model Reconnaissance
Resource Development 3
AML.T0010 ML Supply Chain CompromiseAML.T0019 Publish Poisoned DatasetsAML.T0021 Establish Accounts
Initial Access 3
AML.T0010 ML Supply Chain CompromiseAML.T0011 User ExecutionAML.T0012 Valid Accounts
ML Model Access 2
AML.T0044 Full ML Model AccessAML.T0046 Inference API Access
Execution 6
AML.T0011 User ExecutionAML.T0032 Exposure to Sensitive Adversarial InputsAML.T0051 LLM Prompt InjectionAML.T0051.000 Direct Prompt InjectionAML.T0051.001 Indirect Prompt InjectionAML.T0053 LLM Plugin Compromise
Persistence 1
AML.T0020 Poison Training Data
Privilege Escalation 0
Defense Evasion 3
AML.T0051 LLM Prompt InjectionAML.T0053 LLM Plugin CompromiseAML.T0054 LLM Jailbreak
Credential Access 1
AML.T0012 Valid Accounts
Discovery 1
AML.T0049 Discover ML Artifacts
Collection 3
AML.T0035 ML Artifact CollectionAML.T0043 Model InversionAML.T0057 LLM Data Leakage
ML Attack Staging 2
AML.T0019 Publish Poisoned DatasetsAML.T0020 Poison Training Data
Exfiltration 2
AML.T0024 Exfiltration via ML Inference APIAML.T0057 LLM Data Leakage
Impact 5
AML.T0031 Erode ML Model IntegrityAML.T0032 Exposure to Sensitive Adversarial InputsAML.T0034 Cost HarvestingAML.T0048 External HarmsAML.T0050 Spamming ML System