Detecting AI-generated text (AIGT) remains challenging because existing approaches rely on token-level statistical signals or independent stylometric features, causing them to overfit to specific generators and fail under distribution shift. We identify a structural signal at the sentence-pair level: LLMs produce inter-sentence transition variance that deviates from human writing through inflated variance driven by recurring similarity bursts at paragraph boundaries and templated transitions. We formalize this as Relational Over-Regularization (ROR) and validate it across four benchmarks (p < 0.001). The central contribution is this relational problem formulation, not a novel GNN architecture; CSFG is one concrete instantiation for operationalizing ROR. To exploit this signal, we propose the Cross-Source Stylometric Fingerprint Graph (CSFG), a graph-based framework that encodes positional, sequential, semantic, and transition deviation signals as learnable GNN edge features. The per-edge signed deviation δ_ij operationalizes ROR without hand-crafted thresholds and acts as a false-positive calibrator. CSFG achieves 97.14% accuracy under binary detection, outperforming the strongest graph-based baseline by 11.14 pp, with a false-positive rate of 1.57% and robust generalization to unseen LLMs in the inflated-variance regime; detection degrades for generators whose transition variance falls at or below the human baseline.
Figures & tables
Figure 1: Overview of the CSFG framework.
Dataset
nhuman
nAI
μhuman
μAI
p -value
r
HC3
1000
1000
5.72×10−7
1.05×10−6
3.34×10−45
+0.38
M4
1000
1000
2.52×10−6
3.06×10−6
5.21×10−22
+0.25
MULTITuDE
1000
1000
1.13×10−6
2.12×10−6
9.73×10−60
+0.32
AIGTBench
1000
1000
1.95×10−6
2.84×10−6
1.34×10−73
+0.22
Table 1: Mann–Whitney U test results for Var(T(D)) across four benchmarks. μ denotes the per-group mean variance. r : rank-biserial correlation as effect size. Positive r indicates Var(T(D)) is higher in AI-generated text than in human-written text. All differences are statistically significant ( p<0.001 ).
Figure 2: Distribution of Var(T(D)) for human-written (blue) and AI-generated (orange) text across four benchmarks, as violin plots. The horizontal bar: median. AIGT consistently exhibits higher variance and a heavier upper tail than human-written text, driven by recurring similarity bursts at paragraph boundaries and templated transitions. Significance: ∗∗∗p<0.001 . Effect sizes ( r ) and exact p -values below each panel.
Detector
ACC (%)
FPR (%)
FNR (%)
F-DetectGPT
70.45
17.30
17.30
DNAGPT
81.23
24.60
19.15
LASTDE
58.93
38.82
43.31
Likelhood
60.75
36.96
41.54
DetectAnyLLM
64.56
18.65
52.23
DeTeCtive
87.78
6.74
17.70
Table 2: Binary detection performance under standard evaluation (Setting 1). † RoBERTa is evaluated in an API-only setting via the Hugging Face inference endpoint.
Detector
Generator
Accuracy (%)
FPR (%)
FNR (%)
LASTDE
Claude Sonnet 4.6
50.50
72.00
26.40
GPT-5
51.80
89.00
6.10
Gemini 2.5 Flash Lite
52.30
18.50
77.80
Likelihood
Claude Sonnet 4.6
55.50
68.20
20.10
GPT-5
55.70
37.40
51.40
Gemini 2.5 Flash Lite
52.10
39.00
57.10
Table 3: Detection performance on unseen LLMs (Setting 2). Generators are excluded from training.
Detector
Perturbation
Accuracy (%)
FPR (%)
FNR (%)
LASTDE
Paraphrasing
49.45
19.61
53.77
Humanization
54.16
33.33
47.14
Back-translation
66.73
27.45
33.88
Likelihood
Paraphrasing
67.53
27.45
32.99
Humanization
66.36
19.61
35.10
Back-translation
74.31
25.49
25.71
Table 4: Detection performance under text perturbation (Setting 3) on Claude Sonnet 4.6. Results for GPT-5 and Gemini 2.5 Flash Lite are provided in Appendix D .
Appendix figures & tables12 assets
Supplementary material from the paper’s appendix.
Appendix
λ
Accuracy (%)
FPR (%)
FNR (%)
μ
σ
μ
σ
μ
σ
0.0 (no aux.)
96.25
0.42
2.27
0.31
5.23
0.81
0.1
96.95
0.26
1.68
0.53
4.42
0.83
0.3
97.02
0.24
1.51
0.40
4.44
0.58
0.5
97.19
0.17
1.60
0.38
4.02
0.46
1.0
97.22
0.22
1.48
0.39
4.08
0.45
Appendix
Table 5: Effect of auxiliary loss weight λ on binary detection performance. All results are averaged over 10 runs; σ denotes standard deviation in percentage points (pp). No aux. ( λ=0.0 ) denotes training without the source attribution head. Bold indicates the best value per column.
θ
Accuracy (%)
FPR (%)
FNR (%)
μ
σ
μ
σ
μ
σ
0.3
97.20
0.24
1.48
0.30
4.11
0.55
0.4
97.22
0.21
1.51
0.29
4.05
0.51
0.5
97.14
0.21
1.58
0.36
4.14
0.48
0.6
97.23
0.17
1.53
0.43
4.01
0.45
0.7
97.19
0.22
1.45
0.30
4.16
0.44
Appendix
Table 6: Sensitivity to semantic edge threshold θ . Lower θ yields denser graphs with more long-range semantic edges. All results are averaged over 10 runs; σ denotes standard deviation in percentage points (pp). Bold indicates the best value per column.
Edge Features
Accuracy (%)
FPR (%)
FNR (%)
μ
σ
μ
σ
μ
σ
Full (ours)
97.14
0.27
1.57
0.38
4.15
0.54
w/o sim
96.92
0.21
1.59
0.49
4.57
0.66
w/o dist
96.89
0.24
1.54
0.51
4.67
0.76
w/o is_seq
96.94
0.17
1.69
0.45
4.42
0.61
w/o trans_dev
96.97
0.23
1.65
0.34
4.40
0.67
Appendix
Table 7: Contribution of each edge feature component. Each variant zeros out the corresponding dimension of eij at collation time; graph structure and model architecture remain unchanged. All results are averaged over 10 runs; σ denotes standard deviation in percentage points (pp). Bold indicates the best value per column.
# Layers
Accuracy (%)
FPR (%)
FNR (%)
μ
σ
μ
σ
μ
σ
1
96.59
0.19
2.06
0.31
4.75
0.63
2
97.09
0.26
1.53
0.28
4.29
0.71
3 (ours)
97.15
0.25
1.42
0.43
4.27
0.67
4
97.10
0.31
1.53
0.46
4.26
0.68
Appendix
Table 8: Effect of GNN depth on binary detection performance. All results are averaged over 10 runs; σ denotes standard deviation in percentage points (pp). All variants use the full edge feature vector and default configuration ( θ=0.6 , λ=1.0 ). Bold indicates the best value per column.
Group
n
TV mean
TV std
p -value
KS D
FNR (%)
Human
500
1.1×10−3
6×10−4
–
–
–
GPT-5
500
0.8×10−3
5×10−4
6.05×10−13
0.2389
61.64
Claude Sonnet 4.6
500
1.5×10−3
5×10−4
1.09×10−19
0.2915
20.46
Gemini 2.5 Flash Lite
500
2.3×10−3
6×10−4
8.88×10−102
0.6482
14.81
Appendix
Table 9: Per-generator transition variance statistics on the Setting 2 test set. TV mean and TV std denote the mean and standard deviation of Var(T(D)) per group. p -value: Mann-Whitney U test against the human group. KS D : Kolmogorov-Smirnov distance against the human CDF. FNR: miss rate of CSFG from Table 3.
Figure 3: Cumulative distribution functions of Var(T(D)) for human-written text and three unseen generators (Setting 2 test set). GPT-5 (red) lies to the left of human text (blue), indicating hyper-uniform generation with suppressed transition variance. Gemini 2.5 Flash Lite (green) is displaced far to the right, reflecting an amplified ROR signal. Claude Sonnet 4.6 (orange) occupies an intermediate position. The close overlap between GPT-5 and human CDFs (KS D=0.2389 ) directly explains the elevated FNR reported in Table 3.
Detector
Perturbation
Accuracy (%)
FPR (%)
FNR (%)
LASTDE
Paraphrasing
42.62
9.80
62.32
Humanization
77.72
56.86
18.70
Back-translation
54.14
23.53
48.17
Likelihood
Paraphrasing
60.70
33.33
39.92
Humanization
46.41
21.57
56.91
Back-translation
76.61
27.45
22.97
Appendix
Table 10: Detection performance under text perturbation (Setting 3) on Gemini 2.5 Flash Lite.
Detector
Perturbation
Accuracy (%)
FPR (%)
FNR (%)
LASTDE
Paraphrasing
43.51
22.45
60.00
Humanization
75.62
60.00
20.73
Back-translation
70.11
35.48
29.36
Likelihood
Paraphrasing
65.84
26.53
34.95
Humanization
38.02
15.56
66.74
Back-translation
65.92
16.13
35.78
Appendix
Table 11: Detection performance under text perturbation (Setting 3) on GPT-5.
Generator
N
Acc. (%)
FPR (%)
FNR (%)
Claude Sonnet 4.6
125
56.00
63.86
4.76
Gemini 2.5 Flash Lite
125
51.20
71.08
4.76
GPT-5
125
74.40
18.07
40.48
Appendix
Table 12: Detection performance on documents shorter than 500 characters. FPR increases sharply relative to the 0.66% observed on the full-length Setting 1 test set, while FNR remains low for Claude Sonnet 4.6 and Gemini 2.5 Flash Lite but increases substantially for GPT-5.
Dataset
N
Acc. (%)
FPR (%)
FNR (%)
XSum
100
85.00
25.00
6.00
Appendix
Table 13: CSFG performance on human-written news text (XSum). FPR is substantially elevated relative to the 1.57% observed on the balanced Setting 1 test set.
SeqXGPT
CoCo
CSFG (Ours)
Generator
Perturbation
Acc
FPR
FNR
Acc
FPR
FNR
Acc
FPR
FNR
Claude Sonnet 4.6
Paraphrasing
70.48
9.80
31.57
91.55
84.75
1.67
90.59
88.24
1.22
Humanization
71.90
29.41
27.96
89.22
81.54
4.76
87.43
84.31
5.10
Back-translation
87.43
19.61
11.84
89.55
76.17
3.52
90.96
49.80
4.79
Gemini 2.5 Flash Lite
Paraphrasing
58.30
7.84
45.21
89.92
69.51
3.53
85.98
92.16
5.91
Humanization
64.64
37.25
35.16
79.35
76.17
14.83
76.43
82.35
17.48
Appendix
Table 14: Cross-generator summary of Setting 3 robustness results (%). FPR is measured on perturbed human-written text. Detailed per-generator results are reported separately in Table 4 , 10 , and 11 .
Set.
Condition
Prompt
2
Generation
Please provide a continuation for the following content to make it coherent: { first 30 tokens }
3
Paraphrase
Paraphrase the following text: { Text }
Humanization
Paraphrase the given sentence. Only return the paraphrased sentence in your response. Make it seem like a human wrote the article and that it is from the YOUR SECTION section of YOUR PUBLICATION. Sentence to paraphrase: YOUR SENTENCE Your paraphrase of the sentence: { Text }
Back-translation
You are a professional translator. Translate the following English text into Chinese. Rules: (1) Preserve the original meaning and tone faithfully. (2) Do not add explanations or commentary. (3) Output only the translated Chinese text. Text: { input_text }
You are a professional translator. Translate the following Chinese text back into English. Rules: (1) Translate naturally—do not attempt to recover the original wording. (2) Preserve meaning and tone. (3) Output only the translated English text. Text: { chinese_text }
Appendix
Table 15: Prompts used for text generation (Setting 2) and perturbation (Setting 3).
Recent AI-generated text detection work often introduces a new benchmark together with a specialized detector tailored to it. We revisit this practice from a baseline-first perspective. Across several benchmarks, we show that a plain, fully fine-tuned RoBERTa matches or exceeds the specialized detectors those benchmarks are built around. This suggests that much of the recent architectural complexity is not what drives strong in-distribution detection. The remaining challenge is the distribution shift. The same strong baseline degrades sharply when the topic domain or generating model changes at test time, and simply adding more source data does not close the gap. We identify a key failure mode: under distribution shift, the detector can assign high-confidence machine labels to human-written text from unseen domains. We then study two lightweight domain adaptation methods to address this problem: K-shot adaptation with first-order MAML over LoRA adapters, and a per-sample confidence-weighted ensemble built on top of the adapted detector. Overall, our results suggest that progress in AI-generated text detection should be measured not only by in-distribution performance, but also by robustness under distribution shift.
Zhuoer Shen, Mingyi Wang, Shaofeng Zou +1
Department of Computer Science, University of California, Santa Barbara · School of ECEE, Arizona State University
As large language models (LLMs) generate text that increasingly resembles human writing, the subtle cues that distinguish AI-generated content from human-written content become increasingly challenging to capture. Reliance on generator-specific artifacts is inherently unstable, since new models emerge rapidly and reduce the robustness of such shortcuts. This generalizes unseen generators as a central and challenging problem for AI-text detection. To tackle this challenge, we propose a progressively structured framework that disentangles AI-detection semantics from generator-aware artifacts. This is achieved through a compact latent encoding that encourages semantic minimality, followed by perturbation-based regularization to reduce residual entanglement, and finally a discriminative adaptation stage that aligns representations with task objectives. Experiments on MAGE benchmark, covering 20 representative LLMs across 7 categories, demonstrate consistent improvements over state-of-the-art methods, achieving up to 24.2% accuracy gain and 26.2% F1 improvement. Notably, performance continues to improve as the diversity of training generators increases, confirming strong scalability and generalization in open-set scenarios. Our source code will be publicly available at https://github.com/PuXiao06/DRGD.
Xiao Pu, Zepeng Cheng, Lin Yuan +2
Chongqing University of Posts and Telecommunications, China
AI-generated text (AIGT) detection can be sensitive to the decoding choices of the source large language model (LLM). We observe that perturbing next-token logits or adjusting sampling temperature can reduce detection performance, providing a clear signal of detector vulnerability to decoding-time distribution changes. Building on this observation, we propose EASE (Entropy-Adaptive Distribution Shaping for Evasion), a training-free and detector-agnostic framework for evading AIGT detectors. EASE computes predictive entropy directly from the source LLM's next-token distribution and uses it to adapt both logit perturbation and sampling temperature, without detector feedback or model fine-tuning. Experiments across three source LLMs and multiple detectors demonstrate consistent reductions in detection performance, with negligible degradation in text quality and negligible inference overhead.
Jicheng Zhou, Kahim Wong, Jialong Wang +1
Department of Artificial Intelligence, Faculty of Information Science and Computing State Key Laboratory of Internet of Things for Smart City University of Macau, Macau, China