AML.T0076: Corrupt AI Model
An adversary may purposefully corrupt a malicious AI model file so that it cannot be successfully deserialized in order to evade detection by a model scanner. The corrupt model may still successfully execute malicious code before deserialization fails.
Standards mapping
Where this fits in a TurboPentest engagement
TurboPentest's agentic pentest is powerful and covers a broad range of issues automatically. This particular class is best confirmed in a manual IntegSec engagement, where human pentesters apply deeper methodology and a larger context window than any automated pass.
Frequently asked questions
What is AML.T0076 Corrupt AI Model?
An adversary may purposefully corrupt a malicious AI model file so that it cannot be successfully deserialized in order to evade detection by a model scanner. The corrupt model may still successfully execute malicious code before deserialization fails.
Which tactics does AML.T0076 belong to?
AML.T0076 maps to the Defense Evasion tactic.
Does TurboPentest test for Corrupt AI Model?
TurboPentest's agentic pentest is powerful and covers a broad range of issues automatically. This particular class is best confirmed in a manual IntegSec engagement, where human pentesters apply deeper methodology and a larger context window than any automated pass.
Related MITRE ATLAS techniques
- Initial Access, Defense Evasion, ImpactAML.T0015: Evade AI Model
- Privilege Escalation, Defense EvasionAML.T0054: LLM Jailbreak
- Defense EvasionAML.T0067: LLM Trusted Output Components Manipulation
- Defense EvasionAML.T0068: LLM Prompt Obfuscation
- Defense EvasionAML.T0071: False RAG Entry Injection
- Defense EvasionAML.T0073: Impersonation
Find these issues before an attacker does
TurboPentest runs an agentic AI pentest against your target and reports findings with proof, from $99 per target.
Start a $99 pentest