Search papers, labs, and topics across Lattice.
University of Wisconsin-Madison
5
0
10
After 2022, scientists are not just collaborating more; they鈥檙e venturing into intellectually distant fields, reshaping the landscape of scientific inquiry.
AHA reveals a reusable vulnerability core across production LLM agents, significantly enhancing the efficiency of red-teaming efforts.
VLA models can ace the task but still trigger unsafe outcomes, exposing a critical gap between action execution and semantic understanding.
LLM judges of disinformation risk are internally consistent, but consistently misaligned with actual human readers, raising serious questions about their validity as evaluation proxies.
Prompt leakage attacks on multi-tenant LLMs are far more efficient than previously thought: a new RL-based method reconstructs prompts with over 12x fewer requests.