TRUSTMEM: Learning Trustworthy Memory Consolidation for LLM Agents with Long-Term Memory
For LLM agents requiring reliable long-term memory, TrustMem addresses the critical problem of persistent memory errors that degrade performance over extended interactions.
LLM agents with long-term memory suffer from persistent errors due to unreliable memory updates. TrustMem improves trustworthiness via a Memory Transition Verifier and preference-guided reinforcement learning, achieving state-of-the-art results and reducing omission, corruption, and hallucination by 40.1%, 79.1%, and 50.0% respectively.
Large language model (LLM) agents rely on long-term memory to support extended interactions and personalized assistance beyond finite context windows. Existing memory agents actively update external memory through generated write, revise, and delete operations, but these updates may omit important information, corrupt existing memory, or introduce unsupported hallucinated content. Once stored, such errors become persistent system-state failures that can affect future reasoning and generation. In this paper, we propose TrustMem, a framework designed to improve the trustworthiness of memory consolidation. TrustMem relies on a Memory Transition Verifier to evaluate the transition process of memory updates in terms of coverage, preservation, and faithfulness. It further constructs preference pairs among candidate updates under the same memory state, enabling preference-guided reinforcement learning to directly optimize memory updating behaviors. Extensive experiments demonstrate that TrustMem improves both memory utility and reliability: it achieves state-of-the-art results across MemoryAgentBench, HaluMem, and the Mem-alpha validation set, improves HaluMem memory extraction by 12.14 F1 points, and reduces transition-level omission, corruption, and hallucination by 40.1\%, 79.1\%, and 50.0\%, respectively, compared with the strongest baseline for each error type.