<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="schuster-etal-2020-limitations">
    <titleInfo>
        <title>The Limitations of Stylometry for Detecting Machine-Generated Fake News</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Tal</namePart>
        <namePart type="family">Schuster</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Roei</namePart>
        <namePart type="family">Schuster</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Darsh</namePart>
        <namePart type="given">J</namePart>
        <namePart type="family">Shah</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Regina</namePart>
        <namePart type="family">Barzilay</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2020-06</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <genre authority="bibutilsgt">journal article</genre>
    <relatedItem type="host">
        <titleInfo>
            <title>Computational Linguistics</title>
        </titleInfo>
        <originInfo>
            <issuance>continuing</issuance>
        </originInfo>
        <genre authority="marcgt">periodical</genre>
        <genre authority="bibutilsgt">academic journal</genre>
    </relatedItem>
    <abstract>Recent developments in neural language models (LMs) have raised concerns about their potential misuse for automatically spreading misinformation. In light of these concerns, several studies have proposed to detect machine-generated fake news by capturing their stylistic differences from human-written text. These approaches, broadly termed stylometry, have found success in source attribution and misinformation detection in human-written texts. However, in this work, we show that stylometry is limited against machine-generated misinformation. Whereas humans speak differently when trying to deceive, LMs generate stylistically consistent text, regardless of underlying motive. Thus, though stylometry can successfully prevent impersonation by identifying text provenance, it fails to distinguish legitimate LM applications from those that introduce false information. We create two benchmarks demonstrating the stylistic similarity between malicious and legitimate uses of LMs, utilized in auto-completion and editing-assistance settings.1 Our findings highlight the need for non-stylometry approaches in detecting machine-generated misinformation, and open up the discussion on the desired evaluation benchmarks.</abstract>
    <identifier type="citekey">schuster-etal-2020-limitations</identifier>
    <identifier type="doi">10.1162/coli_a_00380</identifier>
    <location>
        <url>https://aclanthology.org/2020.cl-2.8/</url>
    </location>
    <part>
        <date>2020-06</date>
        <detail type="volume"><number>46</number></detail>
        <detail type="issue"><number>2</number></detail>
        <extent unit="page">
            <start>499</start>
            <end>510</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T The Limitations of Stylometry for Detecting Machine-Generated Fake News
%A Schuster, Tal
%A Schuster, Roei
%A Shah, Darsh J.
%A Barzilay, Regina
%J Computational Linguistics
%D 2020
%8 June
%V 46
%N 2
%F schuster-etal-2020-limitations
%X Recent developments in neural language models (LMs) have raised concerns about their potential misuse for automatically spreading misinformation. In light of these concerns, several studies have proposed to detect machine-generated fake news by capturing their stylistic differences from human-written text. These approaches, broadly termed stylometry, have found success in source attribution and misinformation detection in human-written texts. However, in this work, we show that stylometry is limited against machine-generated misinformation. Whereas humans speak differently when trying to deceive, LMs generate stylistically consistent text, regardless of underlying motive. Thus, though stylometry can successfully prevent impersonation by identifying text provenance, it fails to distinguish legitimate LM applications from those that introduce false information. We create two benchmarks demonstrating the stylistic similarity between malicious and legitimate uses of LMs, utilized in auto-completion and editing-assistance settings.1 Our findings highlight the need for non-stylometry approaches in detecting machine-generated misinformation, and open up the discussion on the desired evaluation benchmarks.
%R 10.1162/coli_a_00380
%U https://aclanthology.org/2020.cl-2.8/
%U https://doi.org/10.1162/coli_a_00380
%P 499-510

Markdown (Informal)

[The Limitations of Stylometry for Detecting Machine-Generated Fake News](https://aclanthology.org/2020.cl-2.8/) (Schuster et al., CL 2020)

• The Limitations of Stylometry for Detecting Machine-Generated Fake News (Schuster et al., CL 2020)

ACL

• Tal Schuster, Roei Schuster, Darsh J. Shah, and Regina Barzilay. 2020. The Limitations of Stylometry for Detecting Machine-Generated Fake News. Computational Linguistics, 46(2):499–510.