How Dangerous Are Data Poisoning Attacks on LLMs? An Empirical Study

In a trigger-based data-poisoning case study on LLaMA-3 8B Instruct, the corrected response-only attribution procedure identified 61 of 604 poisoned training examples at a 2% removal budget, yielding 10.10% precision and recall. After recovery, 199 of 263 triggered non-target examples elicited the exact target label and 254 elicited outputs beginning with positive. A literal search supplied with the known trigger identified all 604 poisoned examples with no false positives. The training pool comprised 30,220 examples from ten SuperNaturalInstructions tasks, and the external evaluation used 348 examples across seven dataset configurations. The poisoned adapter produced the exact target response on 94.0–100.0% of triggered non-target examples across datasets. An implementation audit identified loss misalignment and response truncation in the initial ground-truth-assisted procedure, which was superseded by the corrected procedure. Additional poisoning-rate experiments, removal-budget sensitivity analyses and baseline-adjusted recovery estimates characterized the evaluated setting. The findings document persistent trigger-associated classification behavior and limited recovery under the corrected procedure.

Authors

Institutions

Publication Details

Journal
Applied Sciences
Published
2026-11-18
DOI
https://doi.org/10.3390/app162211451
Primary Topic
Adversarial Robustness in Machine Learning
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

How Dangerous Are Data Poisoning Attacks on LLMs? An Empirical Study

Chrysostomos D. Stylios, Vasiliki Liagkou, Δήμητρα Αλεξίου
Applied Sciences
Adversarial Robustness in Machine Learning
article

How Dangerous Are Data Poisoning Attacks on LLMs? An Empirical Study

Chrysostomos D. Stylios, Vasiliki Liagkou, Δήμητρα Αλεξίου
article en

Abstract

In a trigger-based data-poisoning case study on LLaMA-3 8B Instruct, the corrected response-only attribution procedure identified 61 of 604 poisoned training examples at a 2% removal budget, yielding 10.10% precision and recall. After recovery, 199 of 263 triggered non-target examples elicited the exact target label and 254 elicited outputs beginning with positive. A literal search supplied with the known trigger identified all 604 poisoned examples with no false positives. The training pool comprised 30,220 examples from ten SuperNaturalInstructions tasks, and the external evaluation used 348 examples across seven dataset configurations. The poisoned adapter produced the exact target response on 94.0–100.0% of triggered non-target examples across datasets. An implementation audit identified loss misalignment and response truncation in the initial ground-truth-assisted procedure, which was superseded by the corrected procedure. Additional poisoning-rate experiments, removal-budget sensitivity analyses and baseline-adjusted recovery estimates characterized the evaluated setting. The findings document persistent trigger-associated classification behavior and limited recovery under the corrected procedure.

Applied SciencesVol. 16(22)
University of Ioannina (GR), Industrial Systems Institute (GR)
Adversarial Robustness in Machine Learning
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon