@inproceedings{russo-etal-2026-moraltrajectories,
title = "{M}oral{T}rajectories: Evaluating Moral Steerability in Large Language Models",
author = "Russo, Daniel and
Stranisci, Marco Antonio and
Araque, Oscar and
Guerini, Marco",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.2432/",
pages = "52200--52216",
ISBN = "979-8-89176-462-0",
abstract = "Large language models are increasingly deployed in contexts where contested social and political values are at stake, yet we lack a systematic understanding of how their moral priors shape what they generate. We introduce **{`}MoralTrajectories{`}**, a benchmark for probing morally grounded generation across contentious political and social topics. The benchmark pairs questions from the AllSides Red Blue Translator with first-person belief statements operationalizing the six foundations of Moral Foundations Theory to construct, for each model and topic, a full *moral trajectory*: seven responses spanning an unconstrained baseline and one condition per foundation. We evaluate eight instruction-tuned models across families and sizes using lexical and semantic similarity, distributional divergence, and token-level uncertainty analysis, supported by a human annotation study. Three findings emerge consistently: unconstrained models default to CARE-oriented language, attributable to the empathy-rewarding objectives of instruction tuning; moral prompting shifts generation in foundation-consistent ways, but the effect is predominantly surface-level; and steerability varies substantially across model families and topics. Together, these results suggest that current alignment training collapses moral diversity into a narrow register, and that recovering genuine moral plurality in generation remains an open challenge."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="russo-etal-2026-moraltrajectories">
<titleInfo>
<title>MoralTrajectories: Evaluating Moral Steerability in Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Daniel</namePart>
<namePart type="family">Russo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marco</namePart>
<namePart type="given">Antonio</namePart>
<namePart type="family">Stranisci</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Oscar</namePart>
<namePart type="family">Araque</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marco</namePart>
<namePart type="family">Guerini</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>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Large language models are increasingly deployed in contexts where contested social and political values are at stake, yet we lack a systematic understanding of how their moral priors shape what they generate. We introduce **‘MoralTrajectories‘**, a benchmark for probing morally grounded generation across contentious political and social topics. The benchmark pairs questions from the AllSides Red Blue Translator with first-person belief statements operationalizing the six foundations of Moral Foundations Theory to construct, for each model and topic, a full *moral trajectory*: seven responses spanning an unconstrained baseline and one condition per foundation. We evaluate eight instruction-tuned models across families and sizes using lexical and semantic similarity, distributional divergence, and token-level uncertainty analysis, supported by a human annotation study. Three findings emerge consistently: unconstrained models default to CARE-oriented language, attributable to the empathy-rewarding objectives of instruction tuning; moral prompting shifts generation in foundation-consistent ways, but the effect is predominantly surface-level; and steerability varies substantially across model families and topics. Together, these results suggest that current alignment training collapses moral diversity into a narrow register, and that recovering genuine moral plurality in generation remains an open challenge.</abstract>
<identifier type="citekey">russo-etal-2026-moraltrajectories</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2432/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>52200</start>
<end>52216</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T MoralTrajectories: Evaluating Moral Steerability in Large Language Models
%A Russo, Daniel
%A Stranisci, Marco Antonio
%A Araque, Oscar
%A Guerini, Marco
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F russo-etal-2026-moraltrajectories
%X Large language models are increasingly deployed in contexts where contested social and political values are at stake, yet we lack a systematic understanding of how their moral priors shape what they generate. We introduce **‘MoralTrajectories‘**, a benchmark for probing morally grounded generation across contentious political and social topics. The benchmark pairs questions from the AllSides Red Blue Translator with first-person belief statements operationalizing the six foundations of Moral Foundations Theory to construct, for each model and topic, a full *moral trajectory*: seven responses spanning an unconstrained baseline and one condition per foundation. We evaluate eight instruction-tuned models across families and sizes using lexical and semantic similarity, distributional divergence, and token-level uncertainty analysis, supported by a human annotation study. Three findings emerge consistently: unconstrained models default to CARE-oriented language, attributable to the empathy-rewarding objectives of instruction tuning; moral prompting shifts generation in foundation-consistent ways, but the effect is predominantly surface-level; and steerability varies substantially across model families and topics. Together, these results suggest that current alignment training collapses moral diversity into a narrow register, and that recovering genuine moral plurality in generation remains an open challenge.
%U https://aclanthology.org/2026.findings-emnlp.2432/
%P 52200-52216
Markdown (Informal)
[MoralTrajectories: Evaluating Moral Steerability in Large Language Models](https://aclanthology.org/2026.findings-emnlp.2432/) (Russo et al., Findings 2026)
ACL