PulseAugur
EN
LIVE 07:48:43

LLMs Misled by Office Document Parsing, New Research Reveals

By PulseAugur Editorial · [1 sources] ·

A new research paper highlights a significant security vulnerability in how large language models (LLMs) process office documents. The study reveals that extractors used by LLM applications often misinterpret or omit content from Office Open XML (OOXML) files, leading to discrepancies between what a human sees and what the AI processes. Researchers demonstrated five attacks that successfully misled major LLM assistants like ChatGPT, Claude, Gemini, and Microsoft 365 Copilot, as well as coding agents, by exploiting these extraction divergences without injecting malicious instructions. To address this, a proposed defense mechanism called Prism aims to reconstruct the human-readable view from OOXML specifications before data is fed to the model. AI

IMPACT Highlights a critical security flaw in LLM document ingestion, potentially impacting enterprise adoption and requiring new defense mechanisms.

RANK_REASON Academic paper detailing a security vulnerability in LLM document processing. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

LLMs Misled by Office Document Parsing, New Research Reveals

How we ranked this

Signal score
27 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Academic paper detailing a security vulnerability in LLM document processing. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.AI TIER_1 English(EN) · Penghui Li ·

    When File Ingestion Misleads Language Models: Measuring and Mitigating Divergences in Office Documents

    arXiv:2611.05806v1 Announce Type: cross Abstract: Office documents are now routinely uploaded to large language models (LLMs) for processing. However, an LLM does not always see what a human sees, because it typically reads the text that an extractor produces from the document's …

RELATED ENTITIES

RELATED TOPICS