-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
903 lines (831 loc) · 35.4 KB
/
index.html
File metadata and controls
903 lines (831 loc) · 35.4 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
<!DOCTYPE HTML>
<html>
<head>
<meta charset="utf-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<title>Authorship Attribution in the Era of LLMs: Problems, Methodologies, and Challenges</title>
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="Free HTML5 Website Template by freehtml5.co" />
<meta name="keywords"
content="free website templates, free html5, free template, free bootstrap, free website template, html5, css3, mobile first, responsive" />
<meta name="author" content="freehtml5.co" />
<!-- Begin Jekyll SEO tag v2.8.0 -->
<!-- <meta name="generator" content="Jekyll v3.9.3" /> -->
<meta property="og:title" content="Authorship Attribution in the Era of LLMs" />
<meta property="og:locale" content="en_US" />
<link rel="canonical" href="https://llm-authorship.github.io/" />
<meta property="og:url" content="https://llm-authorship.github.io/" />
<meta property="og:site_name" content="Authorship Attribution in the Era of LLMs" />
<meta property="og:type" content="website" />
<meta property="og:image" content="https://llm-authorship.github.io/static/images/framework.png" />
<meta name="twitter:card" content="summary" />
<meta name="twitter:title" content="Authorship Attribution in the Era of LLMs" />
<meta name="twitter:description" content="Authorship Attribution in the Era of LLMs" />
<meta name="twitter:site" content="@BaixHuang" />
<meta name="twitter:image" content="https://llm-authorship.github.io/static/images/framework.png" />
<script type="text/javascript" async
src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/3.1.0/es5/tex-mml-chtml.js">
</script>
<style>
table {
width: 100%;
border-collapse: collapse;
}
th, td {
border: 1px solid lightgray;
padding: 10px !important;
text-align: justify;
}
th {
background-color: #f2f2f2;
}
.highlight {
background-color: #d9fdd3;
}
.pbox {
display: inline-block;
width: 100%;
}
</style>
<!-- <link href="https://fonts.googleapis.com/css?family=Work+Sans:300,400,500,700,800" rel="stylesheet"> -->
<!-- <link href="https://fonts.googleapis.com/css?family=Inconsolata:400,700" rel="stylesheet"> -->
<!-- Animate.css -->
<link rel="stylesheet" href="css/animate.css">
<!-- Icomoon Icon Fonts-->
<link rel="stylesheet" href="css/icomoon.css">
<!-- Bootstrap -->
<link rel="stylesheet" href="css/bootstrap.css">
<!-- Flexslider -->
<link rel="stylesheet" href="css/flexslider.css">
<!-- <link rel="stylesheet" href="static/css/fontawesome.all.min.css"> -->
<!-- Theme style -->
<link rel="stylesheet" href="css/style.css">
<!-- Modernizr JS -->
<script src="js/modernizr-2.6.2.min.js"></script>
<!-- FOR IE9 below -->
<!--[if lt IE 9]>
<script src="js/respond.min.js"></script>
<![endif]-->
<style>
body {
text-align: justify;
}
</style>
</head>
<body>
<div class="fh5co-loader"></div>
<div id="page">
<nav class="fh5co-nav" role="navigation">
<div class="top-menu">
<div class="container">
<div class="row">
<!-- <div class="col-xs-2">
<div id="fh5co-logo"><a href="index.html"><img src="./static/images/icon.svg" alt="logo" width="50" height="50" style="vertical-align: middle; margin-right: 5px;"></a></div>
</div> -->
<div class="col-xs-10 text-left menu-1">
<ul>
<li class="active"><a href="#home">Home</a></li>
<li><a href="#benchmarks">Benchmarks</a></li>
<li><a href="#detectors">Detectors</a></li>
<li><a href="#canllm-identify-authorship">Can LLMs Identify Authorship (EMNLP)</a></li>
<li><a href="#bibtex">BibTex</a></li>
<li><a href="#contact">Contact</a></li>
</ul>
</div>
</div>
</div>
</div>
</nav>
<!-- Table of Contents -->
<section class="section" id="papers" style="padding: 1rem 1rem 1rem;">
<div class="container is-max-desktop">
<!-- <h2 class="title is-4"">News</h2> -->
<ul style="font-size: 0.95em; line-height: 2; list-style-type: disc; margin-left: 1.2em;">
<li><a href="#home">Who's Your Judge? On the Detectability of LLM-Generated Judgments (New arXiv)</a> [<a href="https://llm-as-a-judge.github.io/" target="_blank">Link</a>]</li>
<li><a href="#home">Authorship Attribution in the Era of LLMs: Problems, Methodologies, and Challenges (SIGKDD Explorations)</a> [<a href="https://github.com/llm-authorship/survey" target="_blank">Link</a>]</li>
<li><a href="#canllm-identify-authorship">Can Large Language Models Identify Authorship (EMNLP)</a> [<a href="https://github.com/baixianghuang/authorship-llm" target="_blank">Link</a>]</li>
</ul>
</div>
</section>
<div style="height: 3rem;"></div>
<header id="fh5co-header" class="fh5co-cover" role="banner">
<div class="overlay"></div>
<div class="container">
<div class="columns">
<div class="col-md-12">
<h1>
<!-- <img src="./static/images/icon.svg" alt="logo" width="42" height="42"> -->
Authorship Attribution in the Era of LLMs: Problems, Methodologies, and Challenges</h1>
<h1 id="home" class="is-size-5 publication-title has-text-justified">
<!-- TLDR: Existing evaluations of knowledge editing often neglect the factual accuracy of LLMs before editing. To address this, HalluEdit offers a comprehensive and faithful benchmark for assessing knowledge editing and other hallucination mitigation methods. -->
</h1>
<div class="is-size-5 publication-authors">
<b><i>ACM SIGKDD Explorations 2024</i></b>
</div>
<!-- <div class="column"></div>
TLDR: This survey paper systematically categorizes authorship attribution in the era of LLMs into four problems: attributing unknown texts to human authors, detecting LLM-generated texts, identifying specific LLMs or human authors, and classifying texts as human-authored, machine-generated, or co-authored by both, while also highlighting key challenges and open problems.
</div> -->
<div>
<span class="author-block"><a href="https://baixianghuang.github.io/">Baixiang Huang</a>,</span>
<span class="author-block"><a href="https://canyuchen.com" target="_blank">Canyu Chen</a>,</span>
<span class="author-block"><a href="https://www.cs.emory.edu/~kshu5/" target="_blank">Kai Shu (Emory University)</a></span>
</div>
<!-- <div class="column">
<span class="author-block">Emory University</span>
</div> -->
<!-- <div class="column author-block">
<span style="color: gray; font-size: smaller; font-style: italic;"><sup>*</sup>Equal
contribution</span>
</div> -->
<div class="column has-text-centered">
<div class="publication-links">
<!-- <span class="link-block">
<a href="https://arxiv.org/abs/"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span class="button-62">Paper</span>
</a>
</span> -->
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2408.08946" target="_blank" class="external-link button is-outlined is-primary">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span class="button-62">arXiv</span>
</a>
</span>
<span class="link-block">
<a href="https://github.com/llm-authorship/survey" target="_blank" class="external-link button is-outlined is-primary">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span class="button-62">Paper List</span>
</a>
</span>
<span class="link-block">
<a href="https://medium.com/@thishuang/authorship-attribution-why-identifying-who-wrote-what-is-more-important-than-ever-in-the-age-of-91fb8cf98096" target="_blank" class="external-link button">
<span class="icon">
<i class="fab fa-medium"></i>
</span>
<span class="button-62">Medium Blog</span>
</a>
</span>
</div>
</div>
</div>
<br>
<div class="column has-text-justified">
<p class="tldr">
<strong>TLDR:</strong>
This survey paper systematically categorizes authorship attribution in the era of LLMs into four problems: attributing unknown texts to human authors, detecting LLM-generated texts, identifying specific LLMs or human authors, and classifying texts as human-authored, machine-generated, or co-authored by both, while also highlighting key challenges and open problems.
</p>
</div>
<!-- <div class="row">
<br>
<div style="text-align: justify;">
<img src="./static/images/framework.png" alt="framework" class="img-responsive" style="width: 80%; margin: 0 auto;">
</div>
</div> -->
</div>
</div>
</div>
</div>
</div>
</div>
</header>
<!-- <script type="module" src="https://gradio.s3-us-west-2.amazonaws.com/4.36.1/gradio.js"></script> -->
<!-- <div class="container">
<br>
<div class="row">
<div class="col-md-12 animate-box">
<gradio-app src="https://boyiwei-cotaeval-leaderboard.hf.space" theme_mode="light"></gradio-app>
</div>
</div>
</div> -->
<br><br>
<section class="section">
<div class="container is-max-desktop" style="text-align: justify;">
<figure style="display: block; text-align: center; margin: 0 auto;">
<img src="./static/images/framework.png" alt="framework" style="max-width: 80%; height: auto;">
<figcaption>
<ol style="text-align: left; display: inline-block;">
Representative Problems in Authorship Attribution:
<li>Human-written Text Attribution (attributing unknown texts to human authors)</li>
<li>LLM-generated Text Detection (detecting if texts are generated by LLMs)</li>
<li>LLM-generated Text Attribution (identifying the specific LLM or human responsible for a text)</li>
<li>Human-LLM Co-authored Text Attribution (classifying texts as human-written, machine-generated, or a combination of both)</li>
</ol>
</figcaption>
</figure>
</div>
<br><br>
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
Accurate attribution of authorship is crucial for maintaining the integrity of digital content, improving forensic investigations, and mitigating the risks of misinformation and plagiarism. Addressing the imperative need for proper authorship attribution is essential to uphold the credibility and accountability of authentic authorship. The rapid advancements of Large Language Models (LLMs) have blurred the lines between human and machine authorship, posing significant challenges for traditional methods. We present a comprehensive literature review that examines the latest research on authorship attribution in the era of LLMs. This survey systematically explores the landscape of this field by categorizing four representative problems: (1) Human-written Text Attribution; (2) LLM-generated Text Detection; (3) LLM-generated Text Attribution; and (4) Human-LLM Co-authored Text Attribution. We also discuss the challenges related to ensuring the generalization and explainability of authorship attribution methods. Generalization requires the ability to generalize across various domains, while explainability emphasizes providing transparent and understandable insights into the decisions made by these models. By evaluating the strengths and limitations of existing methods and benchmarks, we identify key open problems and future research directions in this field. This literature review serves a roadmap for researchers and practitioners interested in understanding the state of the art in this rapidly evolving field.
</p>
</div>
</div>
</div>
</div>
</section>
<section class="section" id="contribution">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3">Contributions</h2>
<div class="content has-text-justified">
<!-- <p>
There's a lot of excellent work that was introduced around the same time as ours.
</p> -->
<!-- <br> -->
<ul>
<li>We provide a timely overview to discuss the challenges and opportunities presented by LLMs in the field of authorship attribution. By systematically categorizing authorship attribution into four problems and balancing problem complexity with practicality, we reveal insights into the evolving filed of authorship attribution in the era of LLMs.</li>
<li>We offer a comprehensive comparison of state-of-the-art methodologies, datasets, benchmarks, and commercial tools used in authorship attribution. This analysis not only improves the understanding of authorship attribution but also provides a valuable resource for researchers and practitioners to use as guidelines for approaching this direction.</li>
<li>We discuss open issues and provide future directions by considering crucial aspects such as generalization, explainability, and interdisciplinary perspectives. We also discuss the broader implications of authorship attribution in real-world applications. This holistic approach ensures that authorship attribution not only yields accurate results but also provides insights that are explainable and socially relevant.</li>
</ul>
<br/>
</div>
</div>
</div>
</div>
</section>
<section class="section" id="benchmarks">
<div class="container is-max-desktop content">
<h2 class="title is-3">Benchmarks</h2>
<p>The table below is a summary of Authorship Attribution Datasets and Benchmarks with LLM-Generated Text. Size is shown as the sum of LLM-generated and human-written texts (with the percentage of human-written texts in parentheses). Language is displayed using the two-letter ISO 639 abbreviation. Columns P2, P3, and P4 indicate whether the dataset supports problems described in Problem 2, 3, and 4, respectively.</p>
<table>
<thead>
<tr>
<th>Name</th>
<th>Domain</th>
<th>Size</th>
<th>Length</th>
<th>Language</th>
<th>Model</th>
<th>P2</th>
<th>P3</th>
<th>P4</th>
</tr>
</thead>
<tbody>
<tr>
<td><a href="https://arxiv.org/pdf/2109.13296">TuringBench</a></td>
<td>News</td>
<td>168,612 (5.2%)</td>
<td>100 to 400 words</td>
<td>en</td>
<td>GPT-1,2,3, GROVER, CTRL, XLM, XLNET, FAIR, TRANSFORMER-XL, PPLM</td>
<td>✓</td>
<td>✓</td>
<td></td>
</tr>
<tr>
<td><a href="https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0251415">TweepFake</a></td>
<td>Social media</td>
<td>25,572 (50.0%)</td>
<td>less than 280 characters</td>
<td>en</td>
<td>GPT-2, RNN, Markov, LSTM, CharRNN</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2304.07666">ArguGPT</a></td>
<td>Academic essays</td>
<td>8,153 (49.5%)</td>
<td>300 words on average</td>
<td>en</td>
<td>GPT2-Xl, text-babbage-001, text-curie-001, davinci-001,002,003, GPT-3.5-Turbo</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2309.11285">AuTexTification</a></td>
<td>Tweets, reviews, news, legal, and how-to articles</td>
<td>163,306 (42.5%)</td>
<td>20 to 100 tokens</td>
<td>en, es</td>
<td>BLOOM, GPT-3</td>
<td>✓</td>
<td>✓</td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/pdf/2304.12008">CHEAT</a></td>
<td>Academic paper abstracts</td>
<td>50,699 (30.4%)</td>
<td>163.9 words on average</td>
<td>en</td>
<td>ChatGPT</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2306.05524">GPABench2</a></td>
<td>Academic paper abstracts</td>
<td>2.385M (6.3%)</td>
<td>70 to 350 words</td>
<td>en</td>
<td>ChatGPT</td>
<td>✓</td>
<td></td>
<td>✓</td>
</tr>
<tr>
<td><a href="https://arxiv.org/pdf/2305.15047">Ghostbuster</a></td>
<td>News, student essays, creative writing</td>
<td>23,091 (87.0%)</td>
<td>77 to 559 (median words per document)</td>
<td>en</td>
<td>ChatGPT, Claude</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2301.07597">HC3</a></td>
<td>Reddit, Wikipedia, medicine, finance</td>
<td>125,230 (64.5%)</td>
<td>25 to 254 words</td>
<td>en, zh</td>
<td>ChatGPT</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2309.02731">HC3 Plus</a></td>
<td>News, social media</td>
<td>214,498</td>
<td>N/A</td>
<td>en, zh</td>
<td>ChatGPT</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/pdf/2310.01307">HC-Var</a></td>
<td>News, reviews, essays, QA</td>
<td>144k (68.8%)</td>
<td>50 to 200 words</td>
<td>en</td>
<td>ChatGPT</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2310.16746">HANSEN</a></td>
<td>Transcripts of speech (spoken text), statements (written text)</td>
<td>535k (96.1%)</td>
<td>less than 1k tokens</td>
<td>en</td>
<td>ChatGPT, PaLM2, Vicuna-13B</td>
<td>✓</td>
<td>✓</td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/pdf/2305.14902">M4</a></td>
<td>Wikipedia, WikiHow, Reddit, QA, news, paper abstracts, peer reviews</td>
<td>147,895 (24.2%)</td>
<td>more than 1k characters</td>
<td>ar, bg, en, id, ru, ur, zh</td>
<td>davinci-003, ChatGPT, GPT-4, Cohere, Dolly2, BLOOMz</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/pdf/2303.14822">MGTBench</a></td>
<td>News, student essays, creative writing</td>
<td>21k (14.3%)</td>
<td>1 to 500 words</td>
<td>en</td>
<td>ChatGPT, ChatGLM, Dolly, GPT4All, StableLM, Claude</td>
<td>✓</td>
<td>✓</td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/pdf/2310.13606">MULTITuDE</a></td>
<td>News</td>
<td>74,081 (10.8%)</td>
<td>200 to 512 tokens</td>
<td>ar, ca, cs, de, en, es, nl, pt, ru, uk, zh</td>
<td>GPT-3,4, ChatGPT, Llama-65B, Alpaca-LoRa-30B, Vicuna-13B, OPT-66B, OPT-IML-Max-1.3B</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/pdf/2305.07969">OpenGPTText</a></td>
<td>OpenWebText</td>
<td>58,790 (50.0%)</td>
<td>less than 2k words</td>
<td>en</td>
<td>ChatGPT</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2311.08723">OpenLLMText</a></td>
<td>OpenWebText</td>
<td>344,530 (20%)</td>
<td>512 tokens</td>
<td>en</td>
<td>ChatGPT, PaLM, Llama, GPT2-XL</td>
<td>✓</td>
<td>✓</td>
<td></td>
</tr>
<tr>
<td><a href="https://aclanthology.org/2023.trustnlp-1.17/">Scientic Paper</a></td>
<td>Scientific papers</td>
<td>29k (55.2%)</td>
<td>900 tokens on average</td>
<td>en</td>
<td>SCIgen, GPT-2,3, ChatGPT, Galactica</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2405.07940">RAID</a></td>
<td>News, Wikipedia, paper abstracts, recipes, Reddit, poems, book summaries, movie reviews</td>
<td>523,985 (2.9%)</td>
<td>323 tokens on average</td>
<td>cs, de, en</td>
<td>GPT-2,3,4, ChatGPT, Mistral-7B, MPT-30B, Llama2-70B, Cohere command and chat</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2402.11175">M4GT-Bench</a></td>
<td>Wikipedia, Wikihow, Reddit, arXiv abstracts, academic paper reviews, student essays</td>
<td>5,368,998 (96.6%)</td>
<td>more than 50 characters</td>
<td>ar, bg, de, en, id, it, ru, ur, zh</td>
<td>ChatGPT, davinci-003, GPT-4, Cohere, Dolly-v2, BLOOMz</td>
<td>✓</td>
<td>✓</td>
<td>✓</td>
</tr>
<tr>
<td><a href="https://arxiv.org/abs/2305.13242">MAGE</a></td>
<td>Reddit, reviews, news, QA, story writing, Wikipedia, academic paper abstracts</td>
<td>448,459 (34.4%)</td>
<td>263 words on average</td>
<td>en</td>
<td>GPT, Llama, GLM-130B, FLAN-T5 OPT, T0, BLOOM-7B1, GPT-J-6B, GPT-NeoX-2</td>
<td>✓</td>
<td></td>
<td></td>
</tr>
<tr>
<td><a href="https://aclanthology.org/2024.findings-naacl.29/">MIXSET</a></td>
<td>Email, news, game reviews, academic paper abstracts, speeches, blogs</td>
<td>3.6k (16.7%)</td>
<td>50 to 250 words</td>
<td>en</td>
<td>GPT-4, Llama2</td>
<td>✓</td>
<td></td>
<td>✓</td>
</tr>
</tbody>
</table>
</div>
</section>
<br>
<section class="section" id="detectors">
<div class="container is-max-desktop content">
<h2 class="title">Detectors</h2>
<p>The Table below presents an overview of LLM-Generated Text Detectors.</p>
<table>
<thead>
<tr>
<th>Detector</th>
<th>Price</th>
<th>API</th>
<th>Website</th>
</tr>
</thead>
<tbody>
<tr>
<td>GPTZero</td>
<td>150k words at $10/month, 10k words for free per month</td>
<td>Yes</td>
<td><a href="https://gptzero.me/">https://gptzero.me/</a></td>
</tr>
<tr>
<td>ZeroGPT</td>
<td>100k characters for $9.99, 15k characters for free</td>
<td>Yes</td>
<td><a href="https://www.zerogpt.com/">https://www.zerogpt.com/</a></td>
</tr>
<tr>
<td>Sapling</td>
<td>50k characters for $25, 2k characters for free</td>
<td>Yes</td>
<td><a href="https://sapling.ai/ai-content-detector">https://sapling.ai/ai-content-detector</a></td>
</tr>
<tr>
<td>Originality.AI</td>
<td>200k words at $14.95/month</td>
<td>Yes</td>
<td><a href="https://originality.ai/">https://originality.ai/</a></td>
</tr>
<tr>
<td>CopyLeaks</td>
<td>300k words at $7.99/month</td>
<td>Yes</td>
<td><a href="https://copyleaks.com/ai-content-detector">https://copyleaks.com/ai-content-detector</a></td>
</tr>
<tr>
<td>Winston</td>
<td>80k words at $12/month</td>
<td>Yes</td>
<td><a href="https://gowinston.ai/">https://gowinston.ai/</a></td>
</tr>
<tr>
<td>GPT Radar</td>
<td>$0.02/100 tokens</td>
<td>N/A</td>
<td><a href="https://gptradar.com/">https://gptradar.com/</a></td>
</tr>
<tr>
<td>Plagiatkontroll</td>
<td>25k words for $15.33/month</td>
<td>Yes</td>
<td><a href="https://plagiatkontroll.no/">https://plagiatkontroll.no/</a></td>
</tr>
<tr>
<td>Turnitin’s AI detector</td>
<td>License required</td>
<td>N/A</td>
<td><a href="https://www.turnitin.com/solutions/topics/ai-writing/ai-detector/">https://www.turnitin.com/solutions/topics/ai-writing/ai-detector/</a></td>
</tr>
<tr>
<td>QuillBot</td>
<td>Free tier: up to 1,200 words; Premium: unlimited ($19.95/month, $99.95/year)</td>
<td>N/A</td>
<td><a href="https://quillbot.com/ai-content-detector">https://quillbot.com/ai-content-detector</a></td>
</tr>
<tr>
<td>GPT-2 Output Detector</td>
<td>Free</td>
<td>N/A</td>
<td><a href="https://github.com/openai/gpt-2-output-dataset/tree/master/detector">https://github.com/openai/gpt-2-output-dataset/tree/master/detector</a></td>
</tr>
<tr>
<td>Crossplag</td>
<td>Free</td>
<td>N/A</td>
<td><a href="https://crossplag.com/ai-content-detector/">https://crossplag.com/ai-content-detector/</a></td>
</tr>
<tr>
<td>CatchGPT</td>
<td>Free</td>
<td>N/A</td>
<td><a href="https://www.catchgpt.ai/">https://www.catchgpt.ai/</a></td>
</tr>
<tr>
<td>Quil.org</td>
<td>Free</td>
<td>N/A</td>
<td><a href="https://aiwritingcheck.org/">https://aiwritingcheck.org/</a></td>
</tr>
<tr>
<td>Scribbr</td>
<td>Free</td>
<td>N/A</td>
<td><a href="https://www.scribbr.com/ai-detector/">https://www.scribbr.com/ai-detector/</a></td>
</tr>
<tr>
<td>Draft Goal</td>
<td>Free</td>
<td>N/A</td>
<td><a href="https://detector.dng.ai/">https://detector.dng.ai/</a></td>
</tr>
<tr>
<td>Writefull</td>
<td>Free</td>
<td>Yes</td>
<td><a href="https://x.writefull.com/gpt-detector">https://x.writefull.com/gpt-detector</a></td>
</tr>
<tr>
<td>Phrasly</td>
<td>Free</td>
<td>Yes</td>
<td><a href="https://phrasly.ai/ai-detector">https://phrasly.ai/ai-detector</a></td>
</tr>
<tr>
<td>Writer</td>
<td>Free</td>
<td>Yes</td>
<td><a href="https://writer.com/ai-content-detector">https://writer.com/ai-content-detector</a></td>
</tr>
</tbody>
</table>
</div>
</section>
<br>
<section class="section" id="bibtex">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code><@article{huang2025authorship,
title={Authorship attribution in the era of llms: Problems, methodologies, and challenges},
author={Huang, Baixiang and Chen, Canyu and Shu, Kai},
journal={ACM SIGKDD Explorations Newsletter},
volume={26},
number={2},
pages={21--43},
year={2025},
publisher={ACM New York, NY, USA}
}</code></pre>
</div>
</section>
<br><br>
<header id="fh5co-header" class="fh5co-cover" role="banner">
<div class="overlay"></div>
<div class="container">
<div class="columns">
<div class="col-md-12">
<h1 id="canllm-identify-authorship">Can Large Language Models Identify Authorship?</h1>
<h1 class="is-size-5 publication-title has-text-justified"></h1>
<div class="is-size-5 publication-authors">
<b><i>EMNLP 2024 Findings</i></b>
</div>
<div>
<span class="author-block"><a href="https://baixianghuang.github.io/">Baixiang Huang</a>,</span>
<span class="author-block"><a href="https://canyuchen.com" target="_blank">Canyu Chen</a>,</span>
<span class="author-block"><a href="https://www.cs.emory.edu/~kshu5/" target="_blank">Kai Shu (Emory University)</a></span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2403.08213" target="_blank" class="external-link button is-outlined is-primary">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span class="button-62">arXiv</span>
</a>
</span>
<span class="link-block">
<a href="https://github.com/baixianghuang/authorship-llm" target="_blank" class="external-link button is-outlined is-primary">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span class="button-62">Data and Code</span>
</a>
</span>
</div>
</div>
</div>
<br>
<div class="column has-text-justified"></div>
<p class="tldr">
<strong>TLDR:</strong>
We propose Linguistically Informed Prompting (LIP) strategy, which offers in-context linguistic guidance, to boost LLMs' reasoning capacity for <i>authorship verification</i> and <i>attribution</i> tasks, while also providing natural language explanations.
</p>
</div>
<!-- <div class="row">
<br>
<img src="./static/images/case.png" alt="case" class="img-responsive">
</div> -->
</div>
</div>
</div>
</div>
</div>
</div>
</header>
<section class="section">
<div class="container is-max-desktop" style="text-align: justify;">
<figure style="display: block; text-align: center; margin: 0 auto;"></figure>
<img src="./static/images/case.png" alt="framework" style="max-width: 100%; height: auto;">
<figcaption>
The example above compares <b>Linguistically Informed Prompting (LIP)</b> with other baselines that provide less linguistic guidance for the authorship verification task. The outputs of an LLM are categorized as either "Analysis" or "Answer." Only the <b>LIP</b> strategy correctly identifies that the two given texts were authored by the same individual. Texts highlighted in <span style="background-color: rgb(255, 204, 168);">orange</span> emphasize the differences across four levels of guidance. Texts highlighted in <span style="background-color: rgb(212, 235, 255);">blue</span> indicate the linguistically informed reasoning process, while <span style="color: rgb(68, 163, 247);">blue</span> text represents content referenced from the original documents.
</figcaption>
</figure>
</div>
<br><br>
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
The ability to accurately identify authorship is crucial for verifying content authenticity and mitigating misinformation. Large Language Models (LLMs) have demonstrated exceptional capacity for reasoning and problem-solving. However, their potential in authorship analysis remains under-explored. Traditional studies have depended on hand-crafted stylistic features, whereas state-of-the-art approaches leverage text embeddings from pre-trained language models. These methods, which typically require fine-tuning on labeled data, often suffer from performance degradation in cross-domain applications and provide limited explainability. This work seeks to address three research questions: (1) Can LLMs perform zero-shot, end-to-end authorship verification effectively? (2) Are LLMs capable of accurately attributing authorship among multiple candidates authors (e.g., 10 and 20)? (3) How can LLMs provide explainability in authorship analysis, particularly through the role of linguistic features? Moreover, we investigate the integration of explicit linguistic features to guide LLMs in their reasoning processes. Our assessment demonstrates LLMs' proficiency in both tasks without the need for domain-specific fine-tuning, providing insights into their decision-making via a detailed analysis of linguistic features. This establishes a new benchmark for future research on LLM-based authorship analysis.
</p>
</div>
</div>
</div>
</div>
</section>
<section class="section" id="contribution">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3">Contributions</h2>
<div class="content has-text-justified">
<ul>
<li>We conduct a comprehensive evaluation of LLMs in authorship attribution and verification tasks. Our results demonstrate that LLMs outperform existing BERT-based models in a zero-shot setting, showcasing their inherent stylometric knowledge essential for distinguishing authorship. This enables them to excel in authorship attribution and verification across low-resource domains without the need for domain-specific fine-tuning.</li>
<li>We develop a pipeline for authorship analysis with LLMs, encompassing dataset preparation, baseline implementation, and evaluation. Our novel Linguistically Informed Prompting (LIP) technique guides LLMs to leverage linguistic features for accurate authorship analysis, enhancing their reasoning capabilities.</li>
<li>Our end-to-end approach improves the explainability of authorship analysis. It elucidates the reasoning and evidence behind authorship predictions, shedding light on how various linguistic features influence these predictions. This contributes to a deeper understanding of the mechanisms behind LLM-based authorship identification.</li>
</ul>
<br/>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code>@inproceedings{huang2024authorship,
title = "Can Large Language Models Identify Authorship?",
author = "Huang, Baixiang and Chen, Canyu and Shu, Kai",
editor = "Al-Onaizan, Yaser and Bansal, Mohit and Chen, Yun-Nung",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2024",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-emnlp.26/",
doi = "10.18653/v1/2024.findings-emnlp.26",
pages = "445--460",
abstract = "The ability to accurately identify authorship is crucial for verifying content authenticity and mitigating misinformation. Large Language Models (LLMs) have demonstrated exceptional capacity for reasoning and problem-solving. However, their potential in authorship analysis remains under-explored. Traditional studies have depended on hand-crafted stylistic features, whereas state-of-the-art approaches leverage text embeddings from pre-trained language models. These methods, which typically require fine-tuning on labeled data, often suffer from performance degradation in cross-domain applications and provide limited explainability. This work seeks to address three research questions: (1) Can LLMs perform zero-shot, end-to-end authorship verification effectively? (2) Are LLMs capable of accurately attributing authorship among multiple candidates authors (e.g., 10 and 20)? (3) Can LLMs provide explainability in authorship analysis, particularly through the role of linguistic features? Moreover, we investigate the integration of explicit linguistic features to guide LLMs in their reasoning processes. Our assessment demonstrates LLMs' proficiency in both tasks without the need for domain-specific fine-tuning, providing explanations into their decision making via a detailed analysis of linguistic features. This establishes a new benchmark for future research on LLM-based authorship analysis."
}</code></pre>
</div>
</section>
<br>
<footer class="footer">
<div class="container">
<div class="columns is-centered">
<div class="column is-8">
<div class="content">
<!-- <p>
This website is licensed under a <a rel="license"
href="http://creativecommons.org/licenses/by-sa/4.0/">Creative
Commons Attribution-ShareAlike 4.0 International License</a>.
</p> -->
<br>
<p id="contact">Contact:
<a href="mailto:bhuang15@hawk.iit.edu">Baixiang Huang</a>, <a href="mailto:kai.shu@emory.edu">Kai Shu</a>
</p>
<p>
</p>
</div>
</div>
</div>
</div>
</footer>
</div>
<div class="gototop js-top">
<a href="#" class="js-gotop"><i class="icon-arrow-up22"></i></a>
</div>
<!-- jQuery -->
<script src="js/jquery.min.js"></script>
<!-- jQuery Easing -->
<script src="js/jquery.easing.1.3.js"></script>
<!-- Bootstrap -->
<script src="js/bootstrap.min.js"></script>
<!-- Waypoints -->
<script src="js/jquery.waypoints.min.js"></script>
<!-- Flexslider -->
<script src="js/jquery.flexslider-min.js"></script>
<!-- Main -->
<script src="js/main.js"></script>
<!-- Default Statcounter code for llm-authorship.github.io
https://llm-authorship.github.io/ -->
<script type="text/javascript">
var sc_project=13027110;
var sc_invisible=1;
var sc_security="8d52e5f6";
</script>
<script type="text/javascript"
src="https://www.statcounter.com/counter/counter.js" async></script>
<noscript><div class="statcounter"><a title="Web Analytics Made Easy -
Statcounter" href="https://statcounter.com/" target="_blank"><img
class="statcounter" src="https://c.statcounter.com/13027110/0/8d52e5f6/1/"
alt="Web Analytics Made Easy - Statcounter"
referrerPolicy="no-referrer-when-downgrade"></a></div></noscript>
<!-- End of Statcounter Code -->
</body>
</html>