Shared long-term memory lets an LLM agent serving one user inadvertently retrieve another user's credentials, but it is unclear when such exposure leads to a…
机构:清华
来源:arXiv 2610.25707 | AI4Papers 论文推荐平台