Scam AI raised $2.6M to combat AI-powered scams→

Log in Request a demo

Research

Publications 25 datasets 14

Datasets sit with the paper that released them; a dataset released without a paper is listed on its own, dated by its version. Downloads are included with any credit purchase on your Scam AI account: one purchase unlocks every dataset open for download, and the credits stay yours to spend on checks. If you use a dataset, please cite both its paper and scam.ai/research, where you got it; the paper's Cite copies both. A dataset without a paper cites its own page. No sign-in needed to cite.

Nov 2026

Version 1.3.1

Document Forgery

Dataset

Genuine PDF Corpus (PDF-Forensics)

82,955 genuine PDFs from 56 public source groups in 14 jurisdictions: legislation and official gazettes, enforcement orders, audit and annual reports, court filings and opinions, regulatory call reports, bank statements, contracts and more. 81,163 are born-digital (10,733 of them digitally signed) and 1,576 are scans. Every file was read CLEAN (no sign of tampering) by our detection engine, release 0.18.0; files it flagged or could not assess were left out. No edited or synthetic documents are included.

Log in to access Dataset page

Citations

This citation references the dataset page Genuine PDF Corpus (PDF-Forensics).

@misc{scamai-pdf-forensics-genuine,
  author       = {{Scam AI}},
  title        = {{Genuine PDF Corpus (PDF-Forensics)}},
  year         = {2026},
  version      = {1.3.1},
  howpublished = {\url{https://scam.ai/research/datasets/pdf-forensics-genuine}},
  note         = {Dataset, version 1.3.1, dated 2026-11-03}
}

Oct 2026

Publication

Can Loan Processors and Underwriters Spot AI-Edited Statements? An Empirical Study

Document Forgery

Citations

This citation references the paper: Can Loan Processors and Underwriters Spot AI-Edited Statements? An Empirical Study.

@misc{ren2026loanprocessorsunderwritersspot,
  title         = {{Can Loan Processors and Underwriters Spot AI-Edited Statements? An Empirical Study}},
  author        = {Simiao Ren and Xingyu Shen and Tommy Duong and Ankit Raj},
  year          = {2026},
  howpublished  = {arXiv:2610.30758},
  eprint        = {2610.30758},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2610.30758}
}

Oct 2026

Publication

Quality-Controlled Reply and Summary Data for Post-Training Small Language Models in AI Call-Screening Agents

Voice & Phone Scams

Citations

This citation references the paper: Quality-Controlled Reply and Summary Data for Post-Training Small Language Models in AI Call-Screening Agents.

@misc{gan2026qualitycontrolledreplysummarydata,
  title         = {{Quality-Controlled Reply and Summary Data for Post-Training Small Language Models in AI Call-Screening Agents}},
  author        = {Jiaqi Gan and Tsang Ng and Ankit Raj and Tommy Duong and Xingyu Shen and Kidus Zewde and Huaiyu Guo and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2610.25458},
  eprint        = {2610.25458},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2610.25458}
}

Oct 2026

Publication

Scam Town: Can Scam Education Spread Like a Vaccine?

Scam Prevention

Citations

This citation references the paper: Scam Town: Can Scam Education Spread Like a Vaccine?.

@misc{ren2026scamtownscameducation,
  title         = {{Scam Town: Can Scam Education Spread Like a Vaccine?}},
  author        = {Simiao Ren and Tommy Duong and Xingyu Shen and Jiaqi Gan and Tsang Ng and Ankit Raj and Kidus Zewde and Huaiyu Guo and Siyu Zhang},
  year          = {2026},
  howpublished  = {arXiv:2610.25040},
  eprint        = {2610.25040},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SI},
  url           = {https://arxiv.org/abs/2610.25040}
}

Sep 2026

Publication

Open-Jev Judgments on CallScreenBench: Calibrated One-Pass Scam Screening with a Small Language Model

Voice & Phone Scams

Dataset

Labeled Scam-Screening Test Calls (CallScreenBench)

82 generated test calls covering 41 CallScreenBench scenarios and two scripted secretaries, with 577 per-turn labels.

Research resource

Citations

This citation references the associated paper: Open-Jev Judgments on CallScreenBench: Calibrated One-Pass Scam Screening with a Small Language Model. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{ren2026openjevjudgmentscallscreenbenchcalibrated,
  title         = {{Open-Jev Judgments on CallScreenBench: Calibrated One-Pass Scam Screening with a Small Language Model}},
  author        = {Simiao Ren and Kidus Zewde and Xingyu Shen and Yuchen Zhou and Dennis Ng and Ankit Raj and Tommy Duong and Yuxin Zhang and Neo Tiangratanakul},
  year          = {2026},
  howpublished  = {arXiv:2609.23959},
  eprint        = {2609.23959},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2609.23959}
}

@misc{scamai-open-jev-callscreenbench,
  author       = {{Scam AI}},
  title        = {{Labeled Scam-Screening Test Calls (CallScreenBench): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-open-jev-callscreenbench}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-open-jev-callscreenbench,
  author       = {{Scam AI}},
  title        = {{Labeled Scam-Screening Test Calls (CallScreenBench): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-open-jev-callscreenbench}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Sep 2026

Publication

Agents That Edit Documents: Measuring Agentic PDF Forgery Against a Non-Agentic Control

Document Forgery

Dataset

AgentForge-Bench

162 PDFs: 81 publicly filed financial documents, each paired with one verified agent-made forgery, plus task cards, the verifier, and per-run verdicts.

Log in to access

Citations

This citation references the associated paper: Agents That Edit Documents: Measuring Agentic PDF Forgery Against a Non-Agentic Control. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{ren2026agentseditdocumentsmeasuring,
  title         = {{Agents That Edit Documents: Measuring Agentic PDF Forgery Against a Non-Agentic Control}},
  author        = {Simiao Ren and Ankit Raj and Tommy Duong and Yuxin Zhang and Dennis Ng and Xingyu Shen and Kidus Zewde and Yuchen Zhou and Neo Tiangratanakul},
  year          = {2026},
  howpublished  = {arXiv:2609.23953},
  eprint        = {2609.23953},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2609.23953}
}

@misc{scamai-agentforge-bench,
  author       = {{Scam AI}},
  title        = {{AgentForge-Bench: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-agentforge-bench}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-agentforge-bench,
  author       = {{Scam AI}},
  title        = {{AgentForge-Bench: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-agentforge-bench}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Sep 2026

Publication

Deepfake Detection Beyond Benchmark Accuracy: Generalization, Robustness, and Reproducibility

Deepfake Detection

Citations

This citation references the paper: Deepfake Detection Beyond Benchmark Accuracy: Generalization, Robustness, and Reproducibility.

@misc{xu2026deepfakedetectionbeyond,
  title         = {{Deepfake Detection Beyond Benchmark Accuracy: Generalization, Robustness, and Reproducibility}},
  author        = {Muduo Xu and Xingyu Shen and Yu Cheng and Ethan Traister and Tommy Duong and Karthik Subramanian and Siyu Zhang and Weidi Xia and Huaiyu Guo and Tyler Wu and Rath Hy and Jiaqi Wu and Yuchen Zhou and Dennis Ng and Simiao Ren},
  year          = {2026},
  howpublished  = {Preprints.org},
  url           = {https://www.preprints.org/manuscript/202609.1394}
}

Sep 2026

Publication

ChatGPT Images 2.5 in the Wild: A Launch-Period Dataset and Detector Evaluation

AI-Generated Detection

Dataset

ChatGPT Images 2.5 in the Wild

3,478 images and 12 videos drawn from 2,440 launch-period posts across eight platforms, with source metadata for detector evaluation.

Log in to access

Citations

This citation references the associated paper: ChatGPT Images 2.5 in the Wild: A Launch-Period Dataset and Detector Evaluation. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{ng2026chatgptimages25wild,
  title         = {{ChatGPT Images 2.5 in the Wild: A Launch-Period Dataset and Detector Evaluation}},
  author        = {Dennis Ng and Xingyu Shen and Ankit Raj and Kidus Zewde and Tommy Duong and Yuchen Zhou and Yuxin Zhang and Neo Tiangratanakul and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2609.15100},
  eprint        = {2609.15100},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2609.15100}
}

@misc{scamai-chatgpt-images-2-5-wild,
  author       = {{Scam AI}},
  title        = {{ChatGPT Images 2.5 in the Wild: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-chatgpt-images-2-5-wild}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-chatgpt-images-2-5-wild,
  author       = {{Scam AI}},
  title        = {{ChatGPT Images 2.5 in the Wild: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-chatgpt-images-2-5-wild}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Sep 2026

Publication

ChatGPT Images 2.5 on Forgery Tasks: Testing Advertised Improvements Against Known Answers

Document Forgery

Dataset

AIForge-Doc v3

A controlled benchmark of ChatGPT Images 2.5 document-forgery tasks, including prompts, masks, scorers, and score files.

Log in to access

Citations

This citation references the associated paper: ChatGPT Images 2.5 on Forgery Tasks: Testing Advertised Improvements Against Known Answers. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{raj2026chatgptimages25forgery,
  title         = {{ChatGPT Images 2.5 on Forgery Tasks: Testing Advertised Improvements Against Known Answers}},
  author        = {Ankit Raj and Yuxin Zhang and Kidus Zewde and Tommy Duong and Jiaqi Gan and Xingyu Shen and Yuchen Zhou and Huaiyu Guo and Siyu Zhang and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2609.13617},
  eprint        = {2609.13617},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2609.13617}
}

@misc{scamai-aiforge-doc-v3,
  author       = {{Scam AI}},
  title        = {{AIForge-Doc v3: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-aiforge-doc-v3}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-aiforge-doc-v3,
  author       = {{Scam AI}},
  title        = {{AIForge-Doc v3: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-aiforge-doc-v3}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Sep 2026

Publication

The Machines Are Calling: Measuring Automated and Synthetic Voices in Unwanted Inbound Calls

Voice & Phone Scams

Dataset

The Machines Are Calling dataset

Derived features and human and detector labels for 6,192 unwanted inbound calls captured by a voice honeypot.

Research resource

Citations

This citation references the associated paper: The Machines Are Calling: Measuring Automated and Synthetic Voices in Unwanted Inbound Calls. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{shen2026machinescallingmeasuringautomated,
  title         = {{The Machines Are Calling: Measuring Automated and Synthetic Voices in Unwanted Inbound Calls}},
  author        = {Xingyu Shen and Tommy Duong and Muduo Xu and Xiaodong An and Jiaqi Gan and Haoyuan Tang and Jamey Z. Liang and Siyu Zhang and Yan Zhang and Ethan Traister and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2609.11137},
  eprint        = {2609.11137},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CR},
  url           = {https://arxiv.org/abs/2609.11137}
}

@misc{scamai-machines-are-calling,
  author       = {{Scam AI}},
  title        = {{The Machines Are Calling dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-machines-are-calling}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-machines-are-calling,
  author       = {{Scam AI}},
  title        = {{The Machines Are Calling dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-machines-are-calling}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Aug 2026

Publication

A Corpus of Real Scam- and Spam-Call Conversations from an Active Voice-Agent Honeypot

Voice & Phone Scams

Citations

This citation references the paper: A Corpus of Real Scam- and Spam-Call Conversations from an Active Voice-Agent Honeypot.

@misc{traister2026corpusrealscamspamcall,
  title         = {{A Corpus of Real Scam- and Spam-Call Conversations from an Active Voice-Agent Honeypot}},
  author        = {Ethan Traister and Dennis Tsang Ng and Siyu Zhang and Huaiyu Guo and Tommy Duong and Tyler Wu and Yuchen Zhou and Xingyu Shen and Jiaqi Wu and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2609.29528},
  eprint        = {2609.29528},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CR},
  url           = {https://arxiv.org/abs/2609.29528}
}

Aug 2026

Publication

Anatomy of a Scam Call: What 10,000 real scam and spam calls reveal about how phone scammers operate

Voice & Phone Scams

Dataset

Real scam & spam call dataset (English)

De-identified English transcripts of 1,000 real scam and spam phone calls, 500 of each, with per-call labels. No audio is included. The open subset of the corpus behind 'Anatomy of a Scam Call'.

Log in to access

Citations

This citation references the associated paper: Anatomy of a Scam Call: What 10,000 real scam and spam calls reveal about how phone scammers operate. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{traister2026anatomyscamcall10000,
  title         = {{Anatomy of a Scam Call: What 10,000 real scam and spam calls reveal about how phone scammers operate}},
  author        = {Ethan Traister and Ankit Raj and Jiaqi Gan and Xingyu Shen and Tyler Wu and Yuchen Zhou and Tommy Duong and Kidus Zewde and Siying Chen and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2608.24127},
  eprint        = {2608.24127},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CR},
  url           = {https://arxiv.org/abs/2608.24127}
}

@misc{scamai-scam-calls-en,
  author       = {{Scam AI}},
  title        = {{Real scam \& spam call dataset (English): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-scam-calls-en}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-scam-calls-en,
  author       = {{Scam AI}},
  title        = {{Real scam \& spam call dataset (English): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-scam-calls-en}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Aug 2026

Publication

CallScreenBench: Benchmarking Small Language Models as Phone Secretaries

Voice & Phone Scams

Citations

This citation references the paper: CallScreenBench: Benchmarking Small Language Models as Phone Secretaries.

@misc{gan2026callscreenbenchbenchmarkingsmalllanguage,
  title         = {{CallScreenBench: Benchmarking Small Language Models as Phone Secretaries}},
  author        = {Jiaqi Gan and Haoyuan Tang and Jamey Z. Liang and Siying Chen and Ankit Raj and Kidus Zewde and Yuchen Zhou and Yuxin Zhang and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2608.01033},
  eprint        = {2608.01033},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CR},
  url           = {https://arxiv.org/abs/2608.01033}
}

Apr 2026

Publication

GPT-Image-2 in the Wild: A Twitter Dataset of Self-Reported AI-Generated Images from the First Week of Deployment

AI-Generated Detection

Dataset

GPT-Image-2 Twitter Dataset

Self-reported AI-generated images collected from Twitter during the first week of GPT-Image-2 deployment — captures real-world distribution shift.

Log in to access

Citations

This citation references the associated paper: GPT-Image-2 in the Wild: A Twitter Dataset of Self-Reported AI-Generated Images from the First Week of Deployment. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{zewde2026gptimage2wildtwitterdataset,
  title         = {{GPT-Image-2 in the Wild: A Twitter Dataset of Self-Reported AI-Generated Images from the First Week of Deployment}},
  author        = {Kidus Zewde and Simiao Ren and Xingyu Shen and Jiaqi Wu and Yuchen Zhou and Tommy Duong and Zikang Zhang and Ethan Traister and Kewen Xie},
  year          = {2026},
  howpublished  = {arXiv:2604.25370},
  eprint        = {2604.25370},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2604.25370}
}

@misc{scamai-gpt-image-2-twitter,
  author       = {{Scam AI}},
  title        = {{GPT-Image-2 Twitter Dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-gpt-image-2-twitter}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-gpt-image-2-twitter,
  author       = {{Scam AI}},
  title        = {{GPT-Image-2 Twitter Dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-gpt-image-2-twitter}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Apr 2026

Publication

When the Forger Is the Judge: GPT-Image-2 Cannot Recognize Its Own Faked Documents

Document Forgery

Dataset

AIForge-Doc v2.0 (GPT-Image-2 document forgeries)

A v2 expansion of AIForge-Doc covering GPT-Image-2 generated tampering — used in the 'When the Forger Is the Judge' benchmark.

Log in to access

Citations

This citation references the associated paper: When the Forger Is the Judge: GPT-Image-2 Cannot Recognize Its Own Faked Documents. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{wu2026forgerjudgegptimage2recognize,
  title         = {{When the Forger Is the Judge: GPT-Image-2 Cannot Recognize Its Own Faked Documents}},
  author        = {Jiaqi Wu and Yuchen Zhou and Dennis Tsang Ng and Xingyu Shen and Kidus Zewde and Ankit Raj and Tommy Duong and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2604.25213},
  eprint        = {2604.25213},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2604.25213}
}

@misc{scamai-aiforge-doc-v2,
  author       = {{Scam AI}},
  title        = {{AIForge-Doc v2.0 (GPT-Image-2 document forgeries): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-aiforge-doc-v2}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-aiforge-doc-v2,
  author       = {{Scam AI}},
  title        = {{AIForge-Doc v2.0 (GPT-Image-2 document forgeries): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-aiforge-doc-v2}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Apr 2026

Publication

A Synthetic Eye Movement Dataset for Script Reading Detection: Real Trajectory Replay on a 3D Simulator

Interview Tech

Dataset

Simulated gaze estimation for reading dataset

Synthetic eye-movement trajectories rendered through a 3D eye simulator, replaying real reading paths — used for script reading detection research.

Log in to access

Citations

This citation references the associated paper: A Synthetic Eye Movement Dataset for Script Reading Detection: Real Trajectory Replay on a 3D Simulator. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{zewde2026syntheticeyemovementdataset,
  title         = {{A Synthetic Eye Movement Dataset for Script Reading Detection: Real Trajectory Replay on a 3D Simulator}},
  author        = {Kidus Zewde and Yuchen Zhou and Dennis Ng and Neo Tiangratanakul and Tommy Duong and Ankit Raj and Yuxin Zhang and Xingyu Shen and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2604.05475},
  eprint        = {2604.05475},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2604.05475}
}

@misc{scamai-synthetic-gaze,
  author       = {{Scam AI}},
  title        = {{Simulated gaze estimation for reading dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-synthetic-gaze}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-synthetic-gaze,
  author       = {{Scam AI}},
  title        = {{Simulated gaze estimation for reading dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-synthetic-gaze}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Apr 2026

Publication

Chinese Language Is Not More Efficient Than English in Vibe Coding: A Preliminary Study on Token Cost and Problem-Solving Rate

LLM Studies

Citations

This citation references the paper: Chinese Language Is Not More Efficient Than English in Vibe Coding: A Preliminary Study on Token Cost and Problem-Solving Rate.

@misc{ren2026chineselanguageefficientenglish,
  title         = {{Chinese Language Is Not More Efficient Than English in Vibe Coding: A Preliminary Study on Token Cost and Problem-Solving Rate}},
  author        = {Simiao Ren and Xingyu Shen and Yuchen Zhou and Dennis and Ng and Ankit Raj},
  year          = {2026},
  howpublished  = {arXiv:2604.14210},
  eprint        = {2604.14210},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2604.14210}
}

Mar 2026

Publication

GPT4o-Receipt: A Dataset and Human Study for AI-Generated Document Forensics

Document Forgery

Dataset

Fully-synthetic AI-generated receipt (GPT-4o-receipt)

Fully synthetic receipts generated by GPT-4o, paired with a human-study evaluation of detectability — used for AI-generated document forensics research.

Log in to access

Citations

This citation references the associated paper: GPT4o-Receipt: A Dataset and Human Study for AI-Generated Document Forensics. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{zhang2026gpt4oreceiptdatasethumanstudy,
  title         = {{GPT4o-Receipt: A Dataset and Human Study for AI-Generated Document Forensics}},
  author        = {Yan Zhang and Simiao Ren and Ankit Raj and En Wei and Dennis Ng and Alex Shen and Jiayu Xue and Yuxin Zhang and Evelyn Marotta},
  year          = {2026},
  howpublished  = {arXiv:2603.11442},
  eprint        = {2603.11442},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2603.11442}
}

@misc{scamai-gpt4o-receipt,
  author       = {{Scam AI}},
  title        = {{Fully-synthetic AI-generated receipt (GPT-4o-receipt): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-gpt4o-receipt}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-gpt4o-receipt,
  author       = {{Scam AI}},
  title        = {{Fully-synthetic AI-generated receipt (GPT-4o-receipt): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-gpt4o-receipt}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Mar 2026

Publication

DOCFORGE-BENCH: A Comprehensive Benchmark for Document Forgery Detection and Analysis

Document Forgery

Citations

This citation references the paper: DOCFORGE-BENCH: A Comprehensive 0-shot Benchmark for Document Forgery Detection and Analysis.

@misc{zhao2026docforgebenchcomprehensive0shotbenchmark,
  title         = {{DOCFORGE-BENCH: A Comprehensive 0-shot Benchmark for Document Forgery Detection and Analysis}},
  author        = {Zengqi Zhao and Weidi Xia and En Wei and Yan Zhang and Jane Mo and Tiannan Zhang and Yuanqin Dai and Zexi Chen and Yiran Tao and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2603.01433},
  eprint        = {2603.01433},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2603.01433}
}

Feb 2026

Publication

AIForge-Doc: A Benchmark for Detecting AI-Forged Tampering in Financial and Form Documents

Document Forgery

Dataset

AI-edit document forgery dataset (AIForge-Doc)

Financial and form documents tampered by a suite of AI editing tools, paired with originals — used to benchmark document forgery detectors.

Log in to access

Citations

This citation references the associated paper: AIForge-Doc: A Benchmark for Detecting AI-Forged Tampering in Financial and Form Documents. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{wu2026aiforgedocbenchmarkdetectingaiforged,
  title         = {{AIForge-Doc: A Benchmark for Detecting AI-Forged Tampering in Financial and Form Documents}},
  author        = {Jiaqi Wu and Yuchen Zhou and Muduo Xu and Zisheng Liang and Simiao Ren and Jiayu Xue and Meige Yang and Siying Chen and Jingheng Huan},
  year          = {2026},
  howpublished  = {arXiv:2602.20569},
  eprint        = {2602.20569},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2602.20569}
}

@misc{scamai-aiforge-doc,
  author       = {{Scam AI}},
  title        = {{AI-edit document forgery dataset (AIForge-Doc): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-aiforge-doc}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-aiforge-doc,
  author       = {{Scam AI}},
  title        = {{AI-edit document forgery dataset (AIForge-Doc): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-aiforge-doc}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Feb 2026

Publication

Can a Teenager Fool an AI? Evaluating Low-Cost Cosmetic Attacks on Age Estimation Systems

Age Estimation

Dataset

Adversarial age estimation attack dataset

Faces with low-cost cosmetic adversarial perturbations designed to defeat age estimation systems — used in 'Can a Teenager Fool an AI?'.

Log in to access

Citations

This citation references the associated paper: Can a Teenager Fool an AI? Evaluating Low-Cost Cosmetic Attacks on Age Estimation Systems. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{shen2026teenagerfoolaievaluating,
  title         = {{Can a Teenager Fool an AI? Evaluating Low-Cost Cosmetic Attacks on Age Estimation Systems}},
  author        = {Xingyu Shen and Tommy Duong and Xiaodong An and Zengqi Zhao and Zebang Hu and Haoyu Hu and Ziyou Wang and Finn Guo and Simiao Ren},
  year          = {2026},
  howpublished  = {arXiv:2602.19539},
  eprint        = {2602.19539},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2602.19539}
}

@misc{scamai-age-adversarial,
  author       = {{Scam AI}},
  title        = {{Adversarial age estimation attack dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-age-adversarial}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-age-adversarial,
  author       = {{Scam AI}},
  title        = {{Adversarial age estimation attack dataset: Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-age-adversarial}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Feb 2026

Publication

How well are open sourced AI-generated image detection models out-of-the-box: A comprehensive benchmark study

AI-Generated Detection

Citations

This citation references the paper: How well are open sourced AI-generated image detection models out-of-the-box: A comprehensive benchmark study.

@misc{ren2026opensourcedaigeneratedimage,
  title         = {{How well are open sourced AI-generated image detection models out-of-the-box: A comprehensive benchmark study}},
  author        = {Simiao Ren and Yuchen Zhou and Xingyu Shen and Kidus Zewde and Tommy Duong and George Huang and Hatsanai and Tiangratanakul and Tsang and Ng and En Wei and Jiayu Xue},
  year          = {2026},
  howpublished  = {arXiv:2602.07814},
  eprint        = {2602.07814},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2602.07814}
}

Feb 2026

Publication

Out of the box age estimation through facial imagery: A Comprehensive Benchmark of Vision-Language Models vs. out-of-the-box Traditional Architectures

Age Estimation

Citations

This citation references the paper: Out of the box age estimation through facial imagery: A Comprehensive Benchmark of Vision-Language Models vs. out-of-the-box Traditional Architectures.

@misc{ren2026boxageestimationfacial,
  title         = {{Out of the box age estimation through facial imagery: A Comprehensive Benchmark of Vision-Language Models vs. out-of-the-box Traditional Architectures}},
  author        = {Simiao Ren and Xingyu Shen and Ankit Raj and Albert Dai and Caroline and Zhang and Yuan Xu and Zexi Chen and Siqi Wu and Chen Gong and Yuxin Zhang},
  year          = {2026},
  howpublished  = {arXiv:2602.07815},
  eprint        = {2602.07815},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2602.07815}
}

Aug 2025

Publication

Can Multi-modal (reasoning) LLMs detect document manipulation?

Document Forgery

Citations

This citation references the paper: Can Multi-modal (reasoning) LLMs detect document manipulation?.

@misc{liang2025multimodalreasoningllmsdetect,
  title         = {{Can Multi-modal (reasoning) LLMs detect document manipulation?}},
  author        = {Zisheng Liang and Kidus Zewde and Rudra Pratap Singh and Disha Patil and Zexi Chen and Jiayu Xue and Yao Yao and Yifei Chen and Qinzhe Liu and Simiao Ren},
  year          = {2025},
  howpublished  = {arXiv:2508.11021},
  eprint        = {2508.11021},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2508.11021}
}

Mar 2025

Publication

Can Multi-modal (reasoning) LLMs work as deepfake detectors?

Deepfake Detection

Citations

This citation references the paper: Can Multi-modal (reasoning) LLMs work as deepfake detectors?.

@misc{ren2025multimodalreasoningllmswork,
  title         = {{Can Multi-modal (reasoning) LLMs work as deepfake detectors?}},
  author        = {Simiao Ren and Yao Yao and Kidus Zewde and Zisheng Liang and Tsang and Ng and Ning-Yau Cheng and Xiaoou Zhan and Qinzhe Liu and Yifei Chen and Hengwei Xu},
  year          = {2025},
  howpublished  = {arXiv:2503.20084},
  eprint        = {2503.20084},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2503.20084}
}

Feb 2025

Publication

Do deepfake detectors work in reality?

Deepfake Detection

Dataset

Real-World Faceswap Dataset (RWFS)

A real-world faceswap collection used to evaluate deepfake detectors against in-the-wild manipulations rather than lab-only synthetics.

Log in to access

Citations

This citation references the associated paper: Do Deepfake Detectors Work in Reality?. If you use the dataset, please cite both the paper and scam.ai/research as the source of the data.

Also cite the dataset source (scam.ai/research)

@misc{ren2025deepfakedetectorsworkreality,
  title         = {{Do Deepfake Detectors Work in Reality?}},
  author        = {Simiao Ren and Hengwei Xu and Tsang Ng and Kidus Zewde and Shengkai Jiang and Ramini Desai and Disha Patil and Ning-Yau Cheng and Yining Zhou and Ragavi Muthukrishnan},
  year          = {2025},
  howpublished  = {arXiv:2502.10920},
  eprint        = {2502.10920},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2502.10920}
}

@misc{scamai-rwfs,
  author       = {{Scam AI}},
  title        = {{Real-World Faceswap Dataset (RWFS): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-rwfs}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

@misc{scamai-rwfs,
  author       = {{Scam AI}},
  title        = {{Real-World Faceswap Dataset (RWFS): Dataset Access Page}},
  year         = {n.d.},
  howpublished = {\url{https://scam.ai/research\#dataset-rwfs}},
  note         = {Dataset obtained from Scam AI Research, https://scam.ai/research}
}

More options

Interested in collaborating?

We partner with academic institutions and industry labs on deepfake detection research.

Propose a collaboration