A Survey on the Security of Long-Term Memory in LLM Agents: Toward Mnemonic Sovereignty

TL;DR

This survey examines the security challenges of long-term, writable memory in LLM agents, emphasizing governance, integrity, and confidentiality issues, and proposing a framework for mnemonic sovereignty.

Contribution

It introduces a memory-lifecycle framework and highlights gaps in current architectures and research on secure, governed persistent memory in LLM agents.

Findings

Research mainly focuses on write and retrieve integrity attacks.

No architecture covers all identified governance primitives.

Use of LLMs for memory security is underexplored but crucial.

Abstract

Research on large language model (LLM) security is shifting from "will the model leak training data" to a more consequential question: can an agent with persistent, long-term memory be continuously shaped, cross-session poisoned, accessed without authorization, and propagated across shared organizational state? Recent surveys cover memory architectures and agent mechanisms, but fewer center the epistemic and governance properties of persistent, writable memory as the reason memory is an independent security problem. This survey addresses that gap. Drawing on cognitive neuroscience and the philosophy of memory, we characterize agent memory as malleable, rewritable, and socially propagating, and develop a memory-lifecycle framework organized around six phases -- Write, Store, Retrieve, Execute, Share, Forget/Rollback -- cross-tabulated against four security objectives: integrity, confidentiality, availability, governance. We organize the literature on memory poisoning, extraction, retrieval corruption, control-flow hijacking, cross-agent propagation, rollback, and governance, and situate representative architectures as determinants of which phases are explicitly governable. Three findings stand out: the literature concentrates on write- and retrieve-time integrity attacks, while confidentiality, availability, store/forget, and benign-persistence failures remain sparsely studied; no published architecture covers all nine governance primitives we identify; and using LLMs themselves for memory security remains sparse yet essential. We unify these under mnemonic sovereignty -- verifiable, recoverable governance over what may be written, who may read, when updates are authorized, and which states may be forgotten -- arguing future secure agents will be differentiated not only by recall capacity, but by memory governance quality.