Poisoning a model's training data plants a hidden backdoor, and the model registry is then the channel that distributes that compromised model into production. Together these two describe one attack chain — from latent data poisoning at pretraining to exploitation once the poisoned model is deployed.
Glossary topic
Model supply-chain poisoning chain
Terms in this topic
Track this in the live feed
See how this plays out in real AI security and governance developments.
Open the feed →