GateMem：多主體共享記憶體代理之記憶體治理基準評測

摘要

大型語言模型代理的記憶基準測試大多假設單一使用者場景，因此醫院、職場、校園及家庭中共同使用的輔助系統仍未被充分研究。在這些部署環境中，多位委託人共同操作一個記憶池，並在不同角色、範疇及關係下進行查詢，因此記憶品質不僅需要檢索能力，更需要治理機制。我們提出 GateMem 基準，針對多委託人共享記憶代理進行評估。該基準同時衡量以下面向：正當長期請求搭配狀態更新的效用、跨情境授權邊界的存取控制，以及明確刪除請求後代理面向的主動遺忘。其涵蓋醫療、辦公、教育及家庭領域，採用長篇幅多方情節、增量式記憶注入、隱藏檢查點、結構化評判及洩漏目標註記。在各種基準方法與基礎模型上，沒有任何方法能同時達成強大效用、穩健存取控制及可靠遺忘。長上下文提示雖能以高 token 成本獲得最佳治理分數，但基於檢索及外部記憶的方法雖降低成本，卻仍會洩漏未經授權或已刪除的資訊。這些結果顯示，目前的記憶代理距離可靠共享機構部署仍有極大差距。

English

Memory benchmarks for LLM agents largely assume single-user settings, leaving shared assistants for hospitals, workplaces, campuses, and households understudied. In these deployments, multiple principals write to a common memory pool and query it under different roles, scopes, and relationships, so memory quality requires governance as well as recall. We introduce GateMem, a benchmark for multi-principal shared-memory agents. GateMem jointly evaluates utility for legitimate long-horizon requests with state updates, access control across contextual authorization boundaries, and agent-facing active forgetting after explicit deletion requests. It spans medical, office, education, and household domains, with long-form multi-party episodes, incremental memory injection, hidden checkpoints, structured judging, and leak-target annotations. Across diverse baselines and backbone models, no method simultaneously achieves strong utility, robust access control, and reliable forgetting. Long-context prompting often yields the best governance score at high token cost, while retrieval-based and external-memory methods reduce cost yet still leak unauthorized or deleted information. These results show current memory agents remain far from reliable shared institutional deployment.