Sheng Zhong
3 indexed papers
Research Timeline
This paper introduces a novel supply-chain attack that uses model code backdoors to actively steal sensitive secrets from local LLM fine-tuning datasets, bypassing current privacy defenses.
This paper introduces UPAttack, a novel threat model demonstrating that focusing on explicit usability requirements can cause LLMs to generate insecure code by neglecting implicit security constraints, and proposes U-SPLOIT to automate this attack.
The authors introduce S$^2$-MAS, a society-scale multi-agent simulation framework that models household decisions and financial behaviors.
Papers
S$^2$-MAS: A Society-Scale Multi-Agent Simulation Framework with 500 Million Household Persona Agents
The authors introduce S$^2$-MAS, a society-scale multi-agent simulation framework that models household decisions and financial behaviors.