Zehua Ma
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Right Tool, Wrong Parameters: Defending LLM Agents Against Parameter Poisoning
According to the abstract, Yao Wang, Zehua Ma, Weiming Zhang and Han Fang study parameter poisoning, a prompt-injection attack that keeps the intended tool call but alters an argument such as a recipient. They propose AuthorityFlow, a defense that traces which information fields may determine each argument.