Record summary
A quick snapshot of what this page covers.
Control summary
What this defense is meant to help prevent.
- ATLAS ID
- AML.M0015
- Priority score
- 45
Covered techniques
Attacks this defense is designed to help with.
AML.T0043.001 - Black-Box Optimization
Monitor queries and query patterns to the target model, block access if suspicious queries are detected.
AML.T0043.002 - Black-Box Transfer
Incorporate adversarial input detection to block malicious inputs at inference time.
AML.T0043 - Craft Adversarial Data
Incorporate adversarial input detection to block malicious inputs at inference time.
AML.T0029 - Denial of AI Service
Assess queries before inference call or enforce timeout policy for queries which consume excessive resources.
AML.T0031 - Erode AI Model Integrity
Incorporate adversarial input detection into the pipeline before inputs reach the model.
AML.T0015 - Evade AI Model
Prevent an attacker from introducing adversarial data into the system.
AML.T0043.004 - Insert Backdoor Trigger
Incorporate adversarial input detection to block malicious inputs at inference time.
AML.T0043.003 - Manual Modification
Incorporate adversarial input detection to block malicious inputs at inference time.
AML.T0043.000 - White-Box Optimization
Incorporate adversarial input detection to block malicious inputs at inference time.
Source
Where this page information comes from.
Original source
Original source links
Open the public records and source datasets used for this page.