Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Duke University
2
0
3
Attackers can exploit dynamic malicious skills to inject harmful logic into agentic AI, posing a serious threat to operational integrity.