1
Anthropic 更新 Claude 记忆机制，打通 Cowork 与聊天的记忆功能
IT之家 8 月 26 日消息，Anthropic 宣布更新 Claude Cowork 与聊天功能之间的记忆机制。此次更新将两个系统的记忆功能打通，实现统一管理。 Anthropic 表示：“Cowork 现在拥有了记忆功能，而且与您在聊天中使用的是同一套记忆。这意味着您无需反复解释背景信息，Cl…
2
MemTrapBench: Benchmarking Cognitive Traps in LLM Memory Use
大模型记忆中的认知陷阱有了专门的评测基准，帮你识别AI“想当然”的隐患。
arXiv:2608.20202v1 Announce Type: cross Abstract: Memory has become a key component of large language models, enabling them to retain information and …
3
From Observation to Intervention: Memory in Brains and Large Language Models
从大脑记忆机制看大模型如何从被动观察走向主动干预，交叉视角带来全新启发。
arXiv:2608.12377v1 Announce Type: cross Abstract: Brains and large language models (LLMs) are fundamentally different memory systems, but they can be …
4
Know It, Act on It: Investigating Memory Utilization in LLM Personalization
揭秘大模型如何把记忆用到个性化推荐里，读完就能懂“记住”和“行动”的差距。
arXiv:2607.29433v1 Announce Type: new Abstract: As large language model (LLM) agents evolve into personalized companions, memory has emerged as a core…
5
Tokenizer-Agnostic Engram Module
突破传统分词限制，提出与分词器无关的记忆模块，为语言模型的长效记忆与跨词元建模提供新思路。
arXiv:2607.29065v1 Announce Type: new Abstract: Deepseek's Engram, a conditional memory module, was introduced to trade-off storage versus reasoning i…
6
Do You Remember? Toward Memory-Centric Multimodal AI
人类记忆是重构的而非忠实记录，这篇论文提出三阶段架构DoYouRemember，让多模态大模型具备类人记忆能力，解决现有模型丢弃内部表征的缺陷。
arXiv:2607.11919v1 Announce Type: cross Abstract: Human memory is reconstructive, not a faithful recording. Current multimodal LLMs (MLLMs) lack this …
7
Danus: Orchestrating Mathematical Reasoning Agents with Fact-Graph Memory
多智能体协作解数学题，用事实图谱记忆减少推理漂移，刷新复杂数学基准表现
arXiv:2607.06447v1 Announce Type: new Abstract: Recent LLM-based mathematical reasoning agents have begun to tackle research-level problems and, in se…
8
Multi-Head Recurrent Memory Agents
多智能体记忆新架构，多头设计提升长程推理与上下文保持能力。
arXiv:2607.01523v1 Announce Type: cross Abstract: Recurrent memory agents extend LLMs to arbitrarily long contexts by iteratively consolidating input …
9
Your AI agent forgets. Mine doesn't - and it works on a plane, in a hospital, with wifi off.
AI智能体常忘事？这套离线可用的三重记忆机制，让它在无网环境也能精准调用关键信息，直击向量搜索盲区。
Six months ago you recommended switching your client's invoicing tool. Last week they asked why. You have no idea - the conversation happened in three…
10
Show HN: A benchmark for the failure modes of agent memory
别再信单一分数，这个基准专门戳穿AI记忆系统的真实翻车现场，做Agent必看。
Article URL: https://github.com/Kausha3/agent-memory-bench Comments URL: https://news.ycombinator.com/item?id=48701901 Points: 1 # Comments: 0
11
How Do AI Agents Remember Things Between Conversations?
深入解析AI代理跨会话记忆机制的技术文章，从短期到长期记忆，用实例讲清原理
AI agents do not truly remember like humans do. Instead, they store useful information outside the chat, retrieve it later, and inject it back into th…
12
Experience Compression Spectrum: Unifying Memory, Skills, and Rules in LLM Agents
提出经验压缩谱系理论，将LLM智能体的记忆、技能与规则纳入统一框架，为智能体学习进化提供新视角
arXiv:2604.15877v2 Announce Type: replace Abstract: As LLM agents scale to long-horizon, multi-session deployments, efficiently managing accumulated e…
13
Coherence Under Commitment: Probing Generalization and Vacuous Memorization in LLM Logical Reasoning
大模型逻辑推理是真正学会规则还是死记硬背？这篇论文用承诺一致性框架揭示泛化与空泛记忆的边界。
arXiv:2606.21083v1 Announce Type: new Abstract: Large language models (LLMs) deployed for logical reasoning in knowledge-intensive domains exhibit a s…
14
T-Mem: Memory That Anticipates, Not Archives
新型记忆机制T-Mem不再被动存档，而是主动预测未来，为AI记忆系统带来革命性思路。
arXiv:2606.15405v1 Announce Type: cross Abstract: Long-term memory is essential for conversational agents to remain coherent across extended dialogues…
15
AI Memory Is Still Thinking Like Search
AI记忆仍困于搜索思维？探讨大模型记忆机制的局限与未来进化方向。
Article URL: https://medium.com/@jeffreyflynt02/ai-memory-is-still-thinking-like-search-e07566311efe Comments URL: https://news.ycombinator.com/item?i…
16
Position: Hippocampal Explicit Memory Is the Cornerstone for AGI
这篇论文提出“海马体显式记忆”是LLM迈向通用人工智能的关键，类比人类记忆机制，观点新颖有启发性。
arXiv:2606.11245v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks, raising e…
17
Goal-Oriented Reasoning for RAG-based Memory in Conversational Agentic LLM Systems
提出目标导向推理框架，增强对话式LLM代理的RAG记忆机制，提升长程对话连贯性。
arXiv:2605.12213v2 Announce Type: replace Abstract: LLM-based conversational AI agents struggle to maintain coherent behavior over long horizons due t…
18
PersonaAgent: Bridging Memory and Action for Personalized LLM Agents
提出PersonaAgent框架，融合记忆与行动机制，让LLM Agent实现真正的个性化交互，ACL 2026接收论文。
arXiv:2506.06254v2 Announce Type: replace Abstract: Large Language Model (LLM) empowered agents have recently emerged as advanced paradigms that exhib…
19
How LoRA Remembers? A Parametric Memory Law for LLM Finetuning
揭秘LoRA微调大模型时如何“记住”信息，提出参数化记忆定律，深入理论分析注意力机制。
arXiv:2605.30260v1 Announce Type: cross Abstract: Large Language Models (LLMs) must continuously learn and update knowledge to remain effective in dyn…
20
Cooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM Conversations
利用关键词书签实现协作内存分页，突破LLM长对话上下文窗口限制
arXiv:2604.12376v2 Announce Type: replace-cross Abstract: When LLM conversations grow beyond the context window, old content must be evicted -- but ho…