Newest papers

Governance-State Corruption in Long-Term Memory for LLM Agents: A Cross-Model Stress Test

Bowen Liu, Youting Wang, Bonan Shen, Dingyan Shang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.30698 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

Un agente LLM puede recuperar una memoria correctamente pero usarla incorrectamente después de que su estado de validez cambie. Estudiamos la corrupción en el estado de gobierno: errores que alteran cuáles registros retenidos entran en el contexto visible del modelo mientras que el contenido almacenado y las tareas permanecen fijos. A través de seis proveedores, 731 bases basadas en referencia producen 2,080 pares de errores coincidentes. En las verificaciones en el plano de control en línea, la corrupción cambia la proyección operativa en el 96.25% de los pares, mientras que la ejecución basada en el oráculo la restaura en el 100%. Para la evaluación a continuación, un panel determinístico seleccionado, no basado en probabilidades, de LongEval/Memora resuelve 298 de los 300 pares de errores coincidentes (238 bases

Machine translation (Meta M2M100); the English original follows.

Abstract

An LLM agent can retrieve a memory correctly yet use it incorrectly after its validity state changes. We study governance-state corruption: faults that alter which retained records enter model-visible context while stored content and tasks remain fixed. Across six providers, 731 benchmark-grounded bases yield 2,080 matched fault pairs. In offline control-plane checks, corruption changes the operative projection in 96.25% of pairs, while oracle-backed enforcement restores it in 100%. For downstream evaluation, a deterministically selected, non-probability LongMemEval/Memora panel resolves 298 of 300 matched triplets (238 distinct bases). Clean-minus-corrupted contrasts are 40.54 percentage points in LongMemEval binary correctness (95% base-cluster interval [30.32, 50.71]) and 30.60 points in rescaled Memora FAMA ([23.31, 37.69]); both sources also show recovery. OAKS Novel native option parity holds for all 6,714 resolved outputs. Full-cohort frozen-scorer corruption/recovery effects (-27.55/+27.51 points) remain descriptive proxies, not validated native semantic population effects. These bounded findings motivate evaluating memory control-state integrity alongside retrieval accuracy.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Correctness, Population, Computer science, Context (archaeology), Econometrics, Psychology

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange