Research by
Hossein Shakibania
012026
VETO
A subtle image cloak that disrupts how modern unified editors attend to a protected reference image.
022026
Obliviate
Guidance-based concept erasure for autoregressive image generators, trained across complete visual-token trajectories.
032026
Token by Token
Backdoor attacks that use ordinary text triggers to jointly manipulate image and language generation in unified autoregressive models.