Search papers, labs, and topics across Lattice.
1
0
3
Single-specialist compromise amplifies harm most strongly in multi-agent LLM systems, highlighting a critical vulnerability beyond simple attack success rates.
