Pinned
Our latest work on model fingerprinting, with @anshulnasery, @alkin_kaz, @sewoong79, and @viswanathpramod, has been accepted to IEEE SaTML 2026.
Fingerprints enable checking model provenance via black-box queries, but most are fully attackable (100% ASR) without even
Are Robust LLM Fingerprints Adversarially Robust?
Our latest paper, accepted to IEEE SaTML 2026, analyzes the robustness of model fingerprinting under adversarial conditions and shows that simple, targeted attacks can reliably defeat many existing fingerprinting strategies.
🧵
