Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models

Joshua Tint

doi:10.18653/v1/2025.queerinai-main.2

Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models

Correct Metadata for

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

Language models have integrated themselves into many aspects of digital life, shaping everything from social media to translation. This paper investigates how large language models (LLMs) respond to LGBTQ+ slang and heteronormative language. Through two experiments, the study assesses the emotional content and the impact of queer slang on responses from models including GPT-3.5, GPT-4o, Llama2, Llama3, Gemma and Mistral. The findings reveal that heteronormative prompts can trigger safety mechanisms, leading to neutral or corrective responses, while LGBTQ+ slang elicits more negative emotions. These insights punctuate the need to provide equitable outcomes for minority slangs and argots, in addition to eliminating explicit bigotry from language models.

Anthology ID:: 2025.queerinai-main.2
Volume:: Proceedings of the Queer in AI Workshop
Month:: May
Year:: 2025
Address:: Hybrid format (in-person and virtual)
Editors:: A Pranav, Alissa Valentine, Shaily Bhatt, Yanan Long, Arjun Subramonian, Amanda Bertsch, Anne Lauscher, Ankush Gupta
Venues:: QueerInAI | WS
SIG:
Publisher:: Association for Computational Linguistics
Note:
Pages:: 6–16
Language:
URL:: https://aclanthology.org/2025.queerinai-main.2/
DOI:: 10.18653/v1/2025.queerinai-main.2
Bibkey:
Cite (ACL):: Joshua Tint. 2025. Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models. In Proceedings of the Queer in AI Workshop, pages 6–16, Hybrid format (in-person and virtual). Association for Computational Linguistics.
Cite (Informal):: Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models (Tint, QueerInAI 2025)
Copy Citation:
PDF:: https://aclanthology.org/2025.queerinai-main.2.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{tint-2025-guardrails,
    title = "Guardrails, not Guidance: Understanding Responses to {LGBTQ}+ Language in Large Language Models",
    author = "Tint, Joshua",
    editor = "Pranav, A  and
      Valentine, Alissa  and
      Bhatt, Shaily  and
      Long, Yanan  and
      Subramonian, Arjun  and
      Bertsch, Amanda  and
      Lauscher, Anne  and
      Gupta, Ankush",
    booktitle = "Proceedings of the Queer in AI Workshop",
    month = may,
    year = "2025",
    address = "Hybrid format (in-person and virtual)",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.queerinai-main.2/",
    doi = "10.18653/v1/2025.queerinai-main.2",
    pages = "6--16",
    ISBN = "979-8-89176-244-2",
    abstract = "Language models have integrated themselves into many aspects of digital life, shaping everything from social media to translation. This paper investigates how large language models (LLMs) respond to LGBTQ+ slang and heteronormative language. Through two experiments, the study assesses the emotional content and the impact of queer slang on responses from models including GPT-3.5, GPT-4o, Llama2, Llama3, Gemma and Mistral. The findings reveal that heteronormative prompts can trigger safety mechanisms, leading to neutral or corrective responses, while LGBTQ+ slang elicits more negative emotions. These insights punctuate the need to provide equitable outcomes for minority slangs and argots, in addition to eliminating explicit bigotry from language models."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tint-2025-guardrails">
    <titleInfo>
        <title>Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Joshua</namePart>
        <namePart type="family">Tint</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-05</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the Queer in AI Workshop</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">A</namePart>
            <namePart type="family">Pranav</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Alissa</namePart>
            <namePart type="family">Valentine</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Shaily</namePart>
            <namePart type="family">Bhatt</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Yanan</namePart>
            <namePart type="family">Long</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Arjun</namePart>
            <namePart type="family">Subramonian</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Amanda</namePart>
            <namePart type="family">Bertsch</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Anne</namePart>
            <namePart type="family">Lauscher</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Ankush</namePart>
            <namePart type="family">Gupta</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Hybrid format (in-person and virtual)</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-244-2</identifier>
    </relatedItem>
    <abstract>Language models have integrated themselves into many aspects of digital life, shaping everything from social media to translation. This paper investigates how large language models (LLMs) respond to LGBTQ+ slang and heteronormative language. Through two experiments, the study assesses the emotional content and the impact of queer slang on responses from models including GPT-3.5, GPT-4o, Llama2, Llama3, Gemma and Mistral. The findings reveal that heteronormative prompts can trigger safety mechanisms, leading to neutral or corrective responses, while LGBTQ+ slang elicits more negative emotions. These insights punctuate the need to provide equitable outcomes for minority slangs and argots, in addition to eliminating explicit bigotry from language models.</abstract>
    <identifier type="citekey">tint-2025-guardrails</identifier>
    <identifier type="doi">10.18653/v1/2025.queerinai-main.2</identifier>
    <location>
        <url>https://aclanthology.org/2025.queerinai-main.2/</url>
    </location>
    <part>
        <date>2025-05</date>
        <extent unit="page">
            <start>6</start>
            <end>16</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models
%A Tint, Joshua
%Y Pranav, A.
%Y Valentine, Alissa
%Y Bhatt, Shaily
%Y Long, Yanan
%Y Subramonian, Arjun
%Y Bertsch, Amanda
%Y Lauscher, Anne
%Y Gupta, Ankush
%S Proceedings of the Queer in AI Workshop
%D 2025
%8 May
%I Association for Computational Linguistics
%C Hybrid format (in-person and virtual)
%@ 979-8-89176-244-2
%F tint-2025-guardrails
%X Language models have integrated themselves into many aspects of digital life, shaping everything from social media to translation. This paper investigates how large language models (LLMs) respond to LGBTQ+ slang and heteronormative language. Through two experiments, the study assesses the emotional content and the impact of queer slang on responses from models including GPT-3.5, GPT-4o, Llama2, Llama3, Gemma and Mistral. The findings reveal that heteronormative prompts can trigger safety mechanisms, leading to neutral or corrective responses, while LGBTQ+ slang elicits more negative emotions. These insights punctuate the need to provide equitable outcomes for minority slangs and argots, in addition to eliminating explicit bigotry from language models.
%R 10.18653/v1/2025.queerinai-main.2
%U https://aclanthology.org/2025.queerinai-main.2/
%U https://doi.org/10.18653/v1/2025.queerinai-main.2
%P 6-16

Download as File

Markdown (Informal)

[Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models](https://aclanthology.org/2025.queerinai-main.2/) (Tint, QueerInAI 2025)

Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models (Tint, QueerInAI 2025)

ACL

Joshua Tint. 2025. Guardrails, not Guidance: Understanding Responses to LGBTQ+ Language in Large Language Models. In Proceedings of the Queer in AI Workshop, pages 6–16, Hybrid format (in-person and virtual). Association for Computational Linguistics.