Modify AI Model Architecture
This is when a hacker physically changes the 'brain' of an AI by adding or removing parts of its structure, similar to rewiring a machine to make it perform different tasks or break down entirely.
This refers to the direct manipulation of a model's computational graph, where an attacker alters the architecture by inserting, deleting, or modifying layers and processing nodes to force unintended behaviors, such as backdoors or performance degradation.
Adversaries may directly modify an AI model's architecture to re-define its behavior. This can include adding or removing layers as well as adding pre or post-processing operations. The effects could include removing the ability to predict certain classes, adding erroneous operations to increase computation costs, or degrading performance. Additionally, a separate adversary-defined network could be injected into the computation graph, which can change the behavior based on the inputs, effectively creating a backdoor.