Esc

Corrupt AI Model - AML.T0076

(ATLAS Technique)


Definition

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.


D3FEND Inferred Relationships





There are no artifacts defined on this offensive technique (yet). Please consider contributing an addition to D3FEND.
json





Model
−
Model
+
Asset Inventory
Asset Vulnerability Enumeration
Container Image Analysis
Configuration Inventory
Data Inventory
Hardware Component Inventory
Network Node Inventory
Software Inventory
Network Mapping
Logical Link Mapping
Active Logical Link Mapping
Passive Logical Link Mapping
Network Traffic Policy Mapping
Network Vulnerability Assessment
Physical Link Mapping
Active Physical Link Mapping
Direct Physical Link Mapping
Operational Activity Mapping
Access Modeling
Operational Dependency Mapping
Operational Risk Assessment
Organization Mapping
System Mapping
Data Exchange Mapping
Service Dependency Mapping
System Dependency Mapping
System Vulnerability Assessment
Harden
−
Harden
+
Agent Authentication
Biometric Authentication
Certificate-based Authentication
Multi-factor Authentication
Password Authentication
Token-based Authentication
Application Hardening
Application Configuration Hardening
Control Flow Integrity
Dead Code Elimination
Exception Handler Pointer Validation
Pointer Authentication
Process Segment Execution Prevention
Segment Address Offset Randomization
Stack Frame Canary Validation
Credential Hardening
Certificate Pinning
Credential Rotation
Certificate Rotation
Password Rotation
One-time Password
Strong Password Policy
Change Default Password
Token Binding
Message Hardening
Message Authentication
Bus Message Authentication
Message Encryption
Transfer Agent Authentication
Platform Hardening
Bootloader Authentication
Disk Encryption
Driver Load Integrity Checking
File Encryption
Hardware-based Write Protection
Physical Enclosure Hardening
Radiation Hardening
Electromagnetic Radiation Hardening
RF Shielding
Software Update
System Configuration Permissions
TPM Boot Integrity
Source Code Hardening
Credential Scrubbing
Domain Logic Validation
Operational Logic Validation
Integer Range Validation
Pointer Validation
Memory Block Start Validation
Null Pointer Checking
Reference Nullification
Trusted Library
Variable Initialization
Variable Type Validation
Detect
−
Detect
+
File Analysis
Dynamic Analysis
Emulated File Analysis
File Content Analysis
File Content Rules
File Hashing
Identifier Analysis
Homoglyph Detection
Identifier Activity Analysis
Identifier Reputation Analysis
Domain Name Reputation Analysis
File Hash Reputation Analysis
IP Reputation Analysis
URL Reputation Analysis
URL Analysis
Message Analysis
Sender MTA Reputation Analysis
Sender Reputation Analysis
Network Traffic Analysis
Administrative Network Activity Analysis
Application Protocol Command Analysis
Remote Firmware Update Monitoring
Byte Sequence Emulation
Certificate Analysis
Active Certificate Analysis
Passive Certificate Analysis
Client-server Payload Profiling
Connection Attempt Analysis
DNS Traffic Analysis
File Carving
Inbound Session Volume Analysis
IPC Traffic Analysis
Network Traffic Community Deviation
Network Traffic Signature Analysis
Per Host Download-Upload Ratio Analysis
Protocol Metadata Anomaly Detection
Relay Pattern Analysis
Remote Terminal Session Detection
RPC Traffic Analysis
Physical Access Monitoring
Electronic Lock Monitoring
Motion Sensor Monitoring
Proximity Sensor Monitoring
Video Surveillance
Platform Monitoring
Application Performance Monitoring
Application Exception Monitoring
File Integrity Monitoring
Firmware Behavior Analysis
Firmware Embedded Monitoring Code
Firmware Verification
Peripheral Firmware Verification
System Firmware Verification
Operating Mode Monitoring
Operating System Monitoring
Endpoint Health Beacon
Input Device Analysis
Memory Boundary Tracking
Scheduled Job Analysis
System Daemon Monitoring
System File Analysis
Service Binary Verification
System Init Config Analysis
User Session Init Config Analysis
Operational Process Monitoring
Platform Uptime Monitoring
Process Analysis
Database Query String Analysis
File Access Pattern Analysis
Indirect Branch Call Analysis
Process Code Segment Verification
Process Self-Modification Detection
Process Spawn Analysis
Process Lineage Analysis
Script Execution Analysis
Shadow Stack Comparisons
System Call Analysis
File Creation Analysis
User Behavior Analysis
Authentication Event Thresholding
Authorization Event Thresholding
Credential Compromise Scope Analysis
Domain Account Monitoring
Job Function Access Pattern Analysis
Local Account Monitoring
Resource Access Pattern Analysis
Session Duration Analysis
User Data Transfer Analysis
User Geolocation Logon Pattern Analysis
Web Session Activity Analysis
Isolate
−
Isolate
+
Access Mediation
Credential Transmission Scoping
IO Port Restriction
Network Access Mediation
LAN Access Mediation
Routing Access Mediation
Network Resource Access Mediation
Remote File Access Mediation
Web Session Access Mediation
Endpoint-based Web Server Access Mediation
Proxy-based Web Server Access Mediation
Operating Mode Restriction
OT Variable Access Restriction
Physical Access Mediation
Physical Locking
System Call Filtering
Local File Access Mediation
Access Policy Administration
Domain Trust Policy
Local File Permissions
User Account Permissions
User Group Permissions
Content Filtering
Content Modification
Content Excision
Content Format Conversion
Content Rebuild
Content Substitution
Content Quarantine
Content Validation
File Format Verification
File Content Decompression Checking
File Internal Structure Verification
File Metadata Consistency Validation
File Metadata Value Verification
File Magic Byte Verification
Execution Isolation
Application-based Process Isolation
Executable Allowlisting
Executable Denylisting
Hardware-based Process Isolation
Kernel-based Process Isolation
Network Isolation
Broadcast Domain Isolation
Directional Network Link
DNS Allowlisting
DNS Denylisting
Forward Resolution Domain Denylisting
Hierarchical Domain Denylisting
Homoglyph Denylisting
Forward Resolution IP Denylisting
Reverse Resolution IP Denylisting
Encrypted Tunnels
Network Traffic Filtering
Inbound Traffic Filtering
Email Filtering
Outbound Traffic Filtering
Deceive
−
Deceive
+
Decoy Environment
Connected Honeynet
Integrated Honeynet
Standalone Honeynet
Decoy Object
Decoy File
Decoy Network Resource
Decoy Persona
Decoy Public Release
Decoy Session Token
Decoy User Credential
Evict
−
Evict
+
Credential Eviction
Account Locking
Authentication Cache Invalidation
Credential Revocation
Object Eviction
Disk Formatting
Disk Erasure
Disk Partitioning
DNS Cache Eviction
Domain Registration Takedown
File Eviction
Email Removal
Registry Key Deletion
Process Eviction
Host Shutdown
Host Reboot
Process Suspension
Process Termination
Session Termination
Restore
−
Restore
+
Restore Access
Reissue Credential
Restore Network Access
Restore User Account Access
Unlock Account
Restore Object
Restore Configuration
Restore Database
Restore Disk Image
Restore File
Restore Email
Restore Software

Use of the MITRE D3FENDâ„¢ Knowledge Graph and website is subject to the Terms of Use. Use of the MITRE D3FEND website is subject to the MITRE D3FEND Privacy Policy. MITRE D3FEND is funded by the National Security Agency (NSA) Cybersecurity Directorate and managed by the National Security Engineering Center (NSEC) which is operated by The MITRE Corporation. MITRE D3FEND; and the MITRE D3FEND logo are trademarks of The MITRE Corporation. This software was produced for the U. S. Government under Basic Contract No. W56KGU-18-D-0004, and is subject to the Rights in Noncommercial Computer Software and Noncommercial Computer Software Documentation Clause 252.227-7014 (FEB 2012)
© 2025 The MITRE Corporation.
Approved for Public Release; Distribution Unlimited #20-2338 and #23-1207.