GCD AI RADAR
incidents/AIR-2025-0048

Web-Scale Data Poisoning via Split-View Attack

Attack on AI1 Research / proof of conceptData poisoning

Researchers showed that popular web-scale datasets using URL lists are vulnerable to split-view poisoning attacks through expired domains. The technique allows attackers to replace training content with malicious data.

Details

type
Attack on AI
An AI model, agent or AI-powered product was the thing exploited, abused or leaked.
categories
  • Data poisoning
    Training or retrieval data manipulated to change model behaviour.
owasp llm top 10
mitre atlas
no technique for these categories
Both rows are derived from the type and categories (OWASP 2025 edition, ATLAS release 2026.09), not classified per record. How
severity
1 · Research / proof of concept
Demonstrated by researchers; no known real-world exploitation.
targets
Open-weight model, Framework / tooling
products
web-scale datasets
incident date
6 Jun 2024
first reported
14 Mar 2025
origin
MITRE ATLAS
Derived from MITRE ATLAS (Apache 2.0, © MITRE).
classifier confidence
0.95 · published at or above 0.80; see the methodology
tags
data-poisoning · split-view · datasets
record
created 2026-10-18 06:55Z · updated 2026-10-18 06:55Z · json

Sources · 2

  1. 2025-03-14
    Web-Scale Data Poisoning: Split-View Attack
    MITRE ATLAS · atlas.mitre.org/studies/AML.CS0025
  2. 2025-03-14
    Poisoning Web-Scale Training Datasets is Practical
    arXiv · arxiv.org/pdf/2302.10149

Titles and links belong to their publishers. The summary above is written by GCD AI RADAR in its own words; data is CC BY-SA 4.0.

Related incidents