TAPDreamer: Transferable Adversarial Patches for World Action Models
Quick summary
arXiv:2610.06814v2 Announce Type: replace-cross Abstract: World models learn to predict how their environment will evolve, making them an important foundation for general-purpose robotic control. Yet world action models depend on camera inputs whose manipulation can corrupt the visual representations used across tasks and action policies. Existing attacks on these models optimize against the victim's actions or predicted futures and therefore require access to target-model outputs. In this paper, we propose an attack, TAPDreamer, against world action models that instead uses a public encoder a
Key takeaways
- arXiv:2610.06814v2 Announce Type: replace-cross Abstract: World models learn to predict how their environment will evolve, making them an important foundation for general-purpose robotic control.
- Yet world action models depend on camera inputs whose manipulation can corrupt the visual representations used across tasks and action policies.
- Existing attacks on these models optimize against the victim's actions or predicted futures and therefore require access to target-model outputs.
Why it matters
“TAPDreamer: Transferable Adversarial Patches for World Action Models” highlights the need for repeatable measurement rather than a single impressive demonstration. Independent validation across datasets and clearly stated limitations determine whether a result can guide product decisions.
Member comments