Search papers, labs, and topics across Lattice.
Duke University
1
0
2
Attackers can exploit dynamic malicious skills to inject harmful logic into agentic AI, posing a serious threat to operational integrity.
