mlsanityData to 14 Oct 2026

MemForm: Benchmarking Memory Formation Poisoning in LLM Agents

arXiv 2610.14431 ↗First posted 10 Oct 2026cs.CR cs.CLcited by 0

On the map: LLM agents and tool use → Agent memory systems

Abstract (arXiv)

Long-term memory enables LLM agents to learn from past interactions but exposes memory formation to adversarial influence. Existing studies primarily organize memory poisoning by attack channels, corruption patterns, or lifecycle stages, leaving formation interfaces insufficiently characterized. We formulate Memory Formation Poisoning as adversarial manipulation of the interfaces through which persistent agent memory is produced and committed, and introduce MemForm, a controlled diagnostic benchmark covering five interfaces: Content, Evidence, Scope, Selection, and Path Poisoning. We introduce Selection and Path Poisoning as controlled intervention categories that isolate retention and consolidation order effects, respectively. Each category pairs clean and poisoned conditions with interface-specific controls. Evaluations across six LLMs and two response formats reveal broad susceptibility at all five interfaces under the evaluated memory protocols and implementations. Crucially, authentic experiences with successful outcomes do not guarantee reliable behavior after memory formation: changes to contextual support, retention, or consolidation order can induce undesirable downstream behavior. In particular, consolidation order can alter behavior even with identical retained experiences, depending on the memory formation procedure. Additional analyses show that interface applicability depends on memory architecture and that Path Poisoning susceptibility depends on the updater. The evaluated defenses provide incomplete mitigation that varies by interface. These findings motivate protecting both what is committed to persistent memory and how memory is formed from experiences. Code and data are available at https://github.com/opanhw/memform-code.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Agent memory systems, one of 16 directions in LLM agents and tool use. Builds on 25 of its 33 references that are in the map. First posted on arXiv 10 Oct 2026.

Builds on

25 of 33 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions7 Jul
  2. 2025Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory28 Apr
  3. 2025A-MEM: Agentic Memory for LLM Agents17 Feb
  4. 2024A Survey on the Memory Mechanism of Large Language Model based Agents21 Apr
  5. 2024Evaluating Very Long-Term Conversational Memory of LLM Agents27 Feb

Other references · 20

  1. 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,568
  2. 2023MemGPT: Towards LLMs as Operating Systemscited by 1,389
  3. 2023AgentBench: Evaluating LLMs as Agentscited by 1,267
  4. 2026DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligencecited by 1,040
  5. 2026GLM-5: from Vibe Coding to Agentic Engineeringcited by 587
  6. 2024InjecAgent: Benchmarking Indirect Prompt Injections in Tool-Integrated Large Language Model Agentscited by 538
  7. 2024AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Basescited by 429
  8. 2024The Task Shield: Enforcing Task Alignment to Defend Against Indirect Prompt Injection in LLM Agentscited by 99
  9. 2025A-MemGuard: A Proactive Defense Framework for LLM-Based Agent Memorycited by 67
  10. 2026Memory Poisoning Attack and Defense on Memory Based LLM-Agentscited by 44
  11. 2026From Untrusted Input to Trusted Memory: A Systematic Study of Memory Poisoning Attacks in LLM Agentscited by 33
  12. 2026A Survey on Long-Term Memory Security in LLM Agents: Attacks, Defenses, and Governance Across the Memory Lifecyclecited by 27
8 more
  1. 2026Hidden in Memory: Sleeper Memory Poisoning in LLM Agentscited by 27
  2. 2026From Storage to Steering: Memory Control Flow Attacks on LLM Agentscited by 19
  3. 2026OEP: Poisoning Self-Evolving LLM Agents via Locally Correct but Non-Transferable Experiencescited by 14
  4. 2026Your Agent's Memories Are Not Its Own: Forged Reasoning Attacks on LLM Agent Memory and Defensescited by 8
  5. 2026MemAudit: Post-hoc Auditing of Poisoned Agent Memory via Causal Attribution and Structural Anomaly Detectioncited by 7
  6. 2026MemPoison: Uncovering Persistent Memory Threats and Structural Blind Spots in LLM Agentscited by 7
  7. 2026MemSecBench: Tracking Agent Memory Poisoning from Persistence to Consequence and Repaircited by 4
  8. 2026SafeGuard: A Multi-Agent Perception-Reasoning Framework for Social-Risk AI-Generated Video Detectioncited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Understanding Stage-Wise Utility-Risk Trade-offs in LLM Agent Memory
  2. 2026Remembering More, Risking More: Longitudinal Safety Risks in Memory-Equipped LLM Agents
  3. 2026MemEvoBench: Benchmarking Safety Risks from Memory Misevolution in LLM Agents
  4. 2026From Attack Success to Attack Severity: Counterfactual Memory Attacks on LLM Agents
  5. 2026State Contamination in Memory-Augmented LLM Agents
  6. 2026MemSentry: A Framework for Detecting Persistent Memory Poisoning in Agentic AI
  7. 2026Salami Attack: Stealthy Collusive Memory Poisoning against OpenClaw
  8. 2026Injection-Execution Dissociation: A Mechanistic Evaluation of Persistent Memory Attacks and Defenses in Stateful LLM Agents