Skip to content

Security & Privacy

Model poisoning

Manipulating a model during training or updates so it behaves maliciously or incorrectly.

Example

Tampered model weights cause a released classifier to behave incorrectly.

Why people use it

Knowing how “Model poisoning” works helps teams protect systems, data, users, and connected tools.

What you'll hear

“What are the risks around Model poisoning if we connect this to company data?”

Related terms