Search papers, labs, and topics across Lattice.
Duke University
3
0
5
Attackers can exploit dynamic malicious skills to inject harmful logic into agentic AI, posing a serious threat to operational integrity.
Vision foundation models are surprisingly brittle: common image edits can drastically alter their embeddings and tank downstream performance.
Finally, a defense against model extraction comes with rigorous theoretical guarantees, using mutual information to certify DNN ownership.
