@inproceedings{semo-etal-2026-imitation,
title = "Imitation Thresholds: Quantifying Influence Propagation in Large Language Models",
author = "Semo, Gil and
Bernsohn, Dor and
Magen, Lior and
Bar, Kfir and
Hayat, Gila",
editor = "Aletras, Nikolaos and
Chalkidis, Ilias and
Barrett, Leslie and
Goanț{\u{a}}, C{\u{a}}t{\u{a}}lina and
Preoțiuc-Pietro, Daniel and
Spanakis, Gerasimos and
Xu, Shanshan",
booktitle = "Proceedings of the Natural Legal Language Processing Workshop 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.nllp-1.33/",
pages = "381--381",
ISBN = "979-8-89176-494-1",
abstract = "As large language models (LLMs) increasingly support sensitive decision-making, understanding their vulnerability to bias becomes critical for maintaining system integrity. However, we lack reliable methods to measure how easily these systems can be compromised through biased training data. We present the first approach to quantifying ``imitation thresholds{``}, defined as the minimum number of training examples needed to significantly shift decision patterns in LLMs applied to legal decision-making. Our experiments span four models across two size categories, 24 judicial personas, and four legal domains, revealing distinct patterns of susceptibility and resistance to persona-driven biases that persist across model architectures and scales. By incrementally fine-tuning LLMs with persona-specific text, we identify thresholds at which model predictions diverge meaningfully from control decision patterns. Our findings show that as few as 80 training examples can reliably alter model behavior, with significant variation across domains and influencer persona types. We further find that susceptibility scales with model capacity, with larger models showing substantially stronger effects, particularly for negative influence. Our work provides a foundational understanding of bias propagation mechanisms in LLMs. We publicly release our datasets and code to support future research in this area."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="semo-etal-2026-imitation">
<titleInfo>
<title>Imitation Thresholds: Quantifying Influence Propagation in Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Gil</namePart>
<namePart type="family">Semo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dor</namePart>
<namePart type="family">Bernsohn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lior</namePart>
<namePart type="family">Magen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kfir</namePart>
<namePart type="family">Bar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gila</namePart>
<namePart type="family">Hayat</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Natural Legal Language Processing Workshop 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nikolaos</namePart>
<namePart type="family">Aletras</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ilias</namePart>
<namePart type="family">Chalkidis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Leslie</namePart>
<namePart type="family">Barrett</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cătălina</namePart>
<namePart type="family">Goanță</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daniel</namePart>
<namePart type="family">Preoțiuc-Pietro</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gerasimos</namePart>
<namePart type="family">Spanakis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shanshan</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-494-1</identifier>
</relatedItem>
<abstract>As large language models (LLMs) increasingly support sensitive decision-making, understanding their vulnerability to bias becomes critical for maintaining system integrity. However, we lack reliable methods to measure how easily these systems can be compromised through biased training data. We present the first approach to quantifying “imitation thresholds“, defined as the minimum number of training examples needed to significantly shift decision patterns in LLMs applied to legal decision-making. Our experiments span four models across two size categories, 24 judicial personas, and four legal domains, revealing distinct patterns of susceptibility and resistance to persona-driven biases that persist across model architectures and scales. By incrementally fine-tuning LLMs with persona-specific text, we identify thresholds at which model predictions diverge meaningfully from control decision patterns. Our findings show that as few as 80 training examples can reliably alter model behavior, with significant variation across domains and influencer persona types. We further find that susceptibility scales with model capacity, with larger models showing substantially stronger effects, particularly for negative influence. Our work provides a foundational understanding of bias propagation mechanisms in LLMs. We publicly release our datasets and code to support future research in this area.</abstract>
<identifier type="citekey">semo-etal-2026-imitation</identifier>
<location>
<url>https://aclanthology.org/2026.nllp-1.33/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>381</start>
<end>381</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Imitation Thresholds: Quantifying Influence Propagation in Large Language Models
%A Semo, Gil
%A Bernsohn, Dor
%A Magen, Lior
%A Bar, Kfir
%A Hayat, Gila
%Y Aletras, Nikolaos
%Y Chalkidis, Ilias
%Y Barrett, Leslie
%Y Goanță, Cătălina
%Y Preoțiuc-Pietro, Daniel
%Y Spanakis, Gerasimos
%Y Xu, Shanshan
%S Proceedings of the Natural Legal Language Processing Workshop 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-494-1
%F semo-etal-2026-imitation
%X As large language models (LLMs) increasingly support sensitive decision-making, understanding their vulnerability to bias becomes critical for maintaining system integrity. However, we lack reliable methods to measure how easily these systems can be compromised through biased training data. We present the first approach to quantifying “imitation thresholds“, defined as the minimum number of training examples needed to significantly shift decision patterns in LLMs applied to legal decision-making. Our experiments span four models across two size categories, 24 judicial personas, and four legal domains, revealing distinct patterns of susceptibility and resistance to persona-driven biases that persist across model architectures and scales. By incrementally fine-tuning LLMs with persona-specific text, we identify thresholds at which model predictions diverge meaningfully from control decision patterns. Our findings show that as few as 80 training examples can reliably alter model behavior, with significant variation across domains and influencer persona types. We further find that susceptibility scales with model capacity, with larger models showing substantially stronger effects, particularly for negative influence. Our work provides a foundational understanding of bias propagation mechanisms in LLMs. We publicly release our datasets and code to support future research in this area.
%U https://aclanthology.org/2026.nllp-1.33/
%P 381-381
Markdown (Informal)
[Imitation Thresholds: Quantifying Influence Propagation in Large Language Models](https://aclanthology.org/2026.nllp-1.33/) (Semo et al., NLLP 2026)
ACL