Jinhao Hu, Bardia Mohammadi, Mofasshara Binte Rafique, Lars Klein, Akhil Arora, Ashvin Goel, Laurent Bindschaedler · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2610.28093 · cited by 0 works (OpenAlex) · open-access PDF
AIエージェントは長期的なメモリを保持し、その多くはウェブページやツールの結果から来ています。エージェントは、モデルを実行するソフトウェアを活用し、このメモリに誰が行動できるかを記録しませんので、収集されたページに隠された指示はユーザー自身の指示として扱うことができます。 ClawMMUは、所有者、範囲、起源、および権限を持つメモリのすべての部分をラベル化する保護層であるClawMMUを提案します。 ClawMMUは、コンテンツがモデルの文脈に入るたびにこれらのラベルをチェックし、ストレージに書かれ、ストレージから戻ってくるか、またはツールの呼び出しを運転します。その中心的なルールは、情報が権威に上が
Machine translation (Meta M2M100); the English original follows.
AI agents keep long-term memory, much of which comes from web pages and tool results. Agent harnesses, the software that runs the model, do not record who may act on this memory, so an instruction hidden in a fetched page can be treated as the user's own instruction. We propose ClawMMU, a protection layer that labels every piece of memory with its owner, scope, origins, and authority. ClawMMU checks these labels whenever content enters the model's context, is written to storage, returns from storage, or drives a tool call. Its central rule is that information may rise in authority, but authority never lifts itself. Because the checks depend on where content came from, not on what it says, memory attacks become classical protection faults that the harness stops even when the model is fooled.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Computer science, Stateful firewall, Software, Layer (electronics), Web page, Programming language
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28093, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange