@inproceedings{senapati-etal-2026-retrieval,
title = "Does Retrieval Augmented Generation Collapse When It Retrieves {AI}-Generated Documents?",
author = "Senapati, Riddhimaan and
Druck, Gregory and
Zeng, Chang and
Fatima, Fabeha and
Rastogi, Rati and
Yilmazel, Ozel",
editor = "Wang, Yimu and
Choi, Yee Man and
Wu, Di and
Ouyang, Siqi and
Azadani, Mozhgan Nasr and
Fung, Yi R. (May)",
booktitle = "Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently ({G}round{LM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.groundlm-1.45/",
pages = "578--605",
ISBN = "979-8-89176-478-1",
abstract = "AI-generated text now constitutes a substantial and growing fraction of indexed web content, creating a feedback loop in which AI assistants may retrieve and condition on their own prior generations. Prior work shows that recursive \textit{training} on model generations causes \textbf{model collapse}: a loss of output diversity and distributional drift. We ask whether a similar collapse emerges at \textit{inference} time, when LLMs retrieve their own generations via RAG without any parameter updates. We measure collapse as the progressive convergence of responses to a narrow set across recursive retrieval rounds. Across three contamination simulations and four open-source LLMs, answer collapse is rapid and consistent under direct contamination, but slower, partial, and model-dependent under realistic competitive retrieval. Collapse can also degrade downstream accuracy on factual multi-hop QA, though the effect is inconsistent across models. We further show that the self-generated documents we introduce are cited far above their share of the retrieved context, an effect that is not fully explained by document quality, writing style, AI authorship, and self-preference. Finally, we propose and evaluate mitigation strategies at both the retrieval and generation stages, none of which reliably prevent collapse. As AI-generated content continues to populate the open web, collapse-resistant retrieval design becomes a prerequisite for trustworthy AI assistants."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="senapati-etal-2026-retrieval">
<titleInfo>
<title>Does Retrieval Augmented Generation Collapse When It Retrieves AI-Generated Documents?</title>
</titleInfo>
<name type="personal">
<namePart type="given">Riddhimaan</namePart>
<namePart type="family">Senapati</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gregory</namePart>
<namePart type="family">Druck</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chang</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fabeha</namePart>
<namePart type="family">Fatima</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rati</namePart>
<namePart type="family">Rastogi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ozel</namePart>
<namePart type="family">Yilmazel</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 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yimu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yee</namePart>
<namePart type="given">Man</namePart>
<namePart type="family">Choi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Di</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siqi</namePart>
<namePart type="family">Ouyang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mozhgan</namePart>
<namePart type="given">Nasr</namePart>
<namePart type="family">Azadani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yi</namePart>
<namePart type="given">R</namePart>
<namePart type="given">(May)</namePart>
<namePart type="family">Fung</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">979-8-89176-478-1</identifier>
</relatedItem>
<abstract>AI-generated text now constitutes a substantial and growing fraction of indexed web content, creating a feedback loop in which AI assistants may retrieve and condition on their own prior generations. Prior work shows that recursive training on model generations causes model collapse: a loss of output diversity and distributional drift. We ask whether a similar collapse emerges at inference time, when LLMs retrieve their own generations via RAG without any parameter updates. We measure collapse as the progressive convergence of responses to a narrow set across recursive retrieval rounds. Across three contamination simulations and four open-source LLMs, answer collapse is rapid and consistent under direct contamination, but slower, partial, and model-dependent under realistic competitive retrieval. Collapse can also degrade downstream accuracy on factual multi-hop QA, though the effect is inconsistent across models. We further show that the self-generated documents we introduce are cited far above their share of the retrieved context, an effect that is not fully explained by document quality, writing style, AI authorship, and self-preference. Finally, we propose and evaluate mitigation strategies at both the retrieval and generation stages, none of which reliably prevent collapse. As AI-generated content continues to populate the open web, collapse-resistant retrieval design becomes a prerequisite for trustworthy AI assistants.</abstract>
<identifier type="citekey">senapati-etal-2026-retrieval</identifier>
<location>
<url>https://aclanthology.org/2026.groundlm-1.45/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>578</start>
<end>605</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Does Retrieval Augmented Generation Collapse When It Retrieves AI-Generated Documents?
%A Senapati, Riddhimaan
%A Druck, Gregory
%A Zeng, Chang
%A Fatima, Fabeha
%A Rastogi, Rati
%A Yilmazel, Ozel
%Y Wang, Yimu
%Y Choi, Yee Man
%Y Wu, Di
%Y Ouyang, Siqi
%Y Azadani, Mozhgan Nasr
%Y Fung, Yi R. (May)
%S Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-478-1
%F senapati-etal-2026-retrieval
%X AI-generated text now constitutes a substantial and growing fraction of indexed web content, creating a feedback loop in which AI assistants may retrieve and condition on their own prior generations. Prior work shows that recursive training on model generations causes model collapse: a loss of output diversity and distributional drift. We ask whether a similar collapse emerges at inference time, when LLMs retrieve their own generations via RAG without any parameter updates. We measure collapse as the progressive convergence of responses to a narrow set across recursive retrieval rounds. Across three contamination simulations and four open-source LLMs, answer collapse is rapid and consistent under direct contamination, but slower, partial, and model-dependent under realistic competitive retrieval. Collapse can also degrade downstream accuracy on factual multi-hop QA, though the effect is inconsistent across models. We further show that the self-generated documents we introduce are cited far above their share of the retrieved context, an effect that is not fully explained by document quality, writing style, AI authorship, and self-preference. Finally, we propose and evaluate mitigation strategies at both the retrieval and generation stages, none of which reliably prevent collapse. As AI-generated content continues to populate the open web, collapse-resistant retrieval design becomes a prerequisite for trustworthy AI assistants.
%U https://aclanthology.org/2026.groundlm-1.45/
%P 578-605
Markdown (Informal)
[Does Retrieval Augmented Generation Collapse When It Retrieves AI-Generated Documents?](https://aclanthology.org/2026.groundlm-1.45/) (Senapati et al., GroundLM 2026)
ACL