Top Deepfake Detection Researchers in AI (2026)

Senior Principal Scientist / Director, Trusted Media Intelligence lab, Adobe Research · Professor of AI; founder-director of DECaDE, University of Surrey (CVSSP / DECaDE)

PhD · University of Bath

Provides technical leadership for Adobe’s Content Authenticity Initiative and the C2PA provenance standard. Developed EKILA attribution.

Co-founded Adobe’s Content Authenticity Initiative in 2019. He provides core technical leadership for C2PA, a cross-industry media-provenance standard. His EKILA work combines robust visual attribution with C2PA to trace provenance for generative art.

Professor, Audio Security and Privacy, EURECOM

Co-founder and lead organizer of the ASVspoof challenges.

Co-founded the ASVspoof and VoicePrivacy community initiatives. He leads EURECOM’s audio security group. Its work on synthetic-speech and replay-attack detection underpins much of modern anti-spoofing research.

Professor, UC Berkeley · Co-Founder & Chief Science Officer, GetReal Security

PhD · University of Pennsylvania — PhD in Computer Science (1997), advised by Eero Simoncelli/David Brainard-era vision group; postdoc at MIT with Ted Adelson

Pioneered digital image forensics and deepfake detection. UC Berkeley professor and GetReal Security co-founder.

Founded modern digital image forensics through two decades of work on detecting manipulated and AI-generated media. He co-founded GetReal Security and is one of the most-quoted experts on deepfakes and authenticity.

Associate Director, Barcelona Supercomputing Center (BSC-CNS)

PhD · Universitat Politècnica de Catalunya (Technical University of Catalonia) — PhD (2009), Image Processing Group; dissertation 'Human Motion Capture with Scalable Body Models'

Led Meta’s AI Red Team and the Deepfake Detection Challenge (DFDC).

Led the DeepFake Detection Challenge (DFDC) at Meta, the largest public deepfake-detection competition and dataset. He works on responsible AI and forensics datasets.

Professor of Speech Technology, University of Eastern Finland

PhD · University of Joensuu

Co-founded the ASVspoof challenges. Coauthored t-DCF, an evaluation metric that accounts for spoofing.

Co-founded ASVspoof, the field-defining benchmark for speech anti-spoofing. He co-developed the tandem detection cost function (t-DCF). It became the standard metric for jointly evaluating spoofing countermeasures and speaker verification.

SUNY Distinguished Professor & Empire Innovation Professor; Director, Institute for AI and Data Science, University at Buffalo (SUNY)

PhD · Dartmouth College — PhD in Computer Science (2005), advised by Hany Farid

Leads media-forensics and deepfake-detection research at the University at Buffalo. Built DeepFake-o-Meter.

Created DeepfakeBench and the public Deepfake-o-meter detection platform. He pioneered early detection methods that use physiological cues such as eye blinking.

Associate Professor of Computer Science, Cornell Tech (Cornell University)

PhD · MIT — Advised by William T. Freeman and Antonio Torralba (CSAIL)

Studies audio-visual self-supervised learning and self-supervised video forensics for deepfake detection.

Developed widely cited methods for universal detection of CNN- and diffusion-generated images and for audio-visual forensics at the University of Michigan.

Full Professor, University of Naples Federico II

PhD · University of Naples Federico II — Multimedia/image processing group, Dept. of Electrical Engineering and Information Technology (now leads the GRIP Multimedia Forensics Lab)

Leads multimedia-forensics research at the University of Naples Federico II. Studies synthetic-image and deepfake detection in the GRIP lab.

Co-created FaceForensics++, the most widely used benchmark for face-manipulation detection. She is a leading authority on image and video forensics and GAN-generated-content detection.

Professor, National Institute of Informatics (NII)

PhD · Tokyo Institute of Technology — PhD in Information Processing (2006), Interdisciplinary Graduate School of Science and Engineering (Tokuda/Kobayashi speech synthesis group)

Pioneered speech synthesis and audio deepfake anti-spoofing at NII Tokyo. Co-leads the ASVspoof challenge.

Co-founded the ASVspoof challenges, the global standard for audio deepfake and voice-clone detection (anti-spoofing). He also leads speech-synthesis research.

Professor of Electrical and Computer Engineering, Clemson University

PhD · University of New Mexico — PhD in computer engineering (with Distinction), 2003

Studies deepfakes and media forensics, including recurrent-CNN and two-branch video detection. Previously at USC ISI, now at Clemson.

A leading researcher in deepfake detection, biometric anti-spoofing, and multimedia forensics.

Associate Professor, ISPL, Politecnico di Milano

PhD · Politecnico di Milano — Image and Sound Processing Lab (ISPL)

Studies image, video, and audio forensics in the ISPL lab. Contributes to DARPA MediFor and SemaFor.

Associate Professor in Politecnico di Milano’s ISPL group and Associate Editor of IEEE TIFS. He is co-PI on DARPA’s MediFor and SemaFor programs. His work covers image, video, and audio deepfake forensics.

Research Scientist, Anthropic

PhD · UC Berkeley — Advised by David Wagner (security/adversarial ML)

Studies adversarial ML and ML security. Created the Carlini & Wagner attack and methods for extracting LLM training data.

A leading adversarial-ML researcher. He tests watermark and deepfake-detector robustness against attacks, helping define what detection can withstand.

Tenure-Track Assistant Professor (GRIP group), University of Naples Federico II

PhD · University of Naples Federico II — GRIP (Image Processing Research Group), advised by Luisa Verdoliva

Studies media forensics and deepfake detection at GRIP, University of Naples. Coauthored FaceForensics++ and ID-Reveal.

Coauthored FaceForensics++. He leads research on splice and copy-move detection and forensic fingerprints of generative models, including Noiseprint.

Research Scientist, Google DeepMind

PhD · California Institute of Technology — Richard M. Murray

Lead author of SynthID-Text, the first production-scale LLM text-watermarking scheme (Nature 2024).

Lead author of the Nature 2024 paper that introduced SynthID-Text. This production-ready watermarking scheme preserves text quality with minimal latency. Google DeepMind deployed it and released it as open source.

Research Scientist, Microsoft

Coauthored t-DCF and contributed ASVspoof evaluation infrastructure.

Coauthored t-DCF, an evaluation metric that accounts for spoofing. At EURECOM, he made key contributions to ASVspoof metrics and countermeasures. He now works as a research scientist at Microsoft, previously Nuance.

Charles William Harrison Distinguished Professor of Electrical and Computer Engineering, Purdue University

PhD · Purdue University — PhD in electrical engineering, Purdue University

Directs Purdue’s VIPER Lab. Leads DARPA SemaFor media-forensics work on deepfakes and manipulated media.

A long-standing media-forensics leader. He played a central role in DARPA’s MediFor program for image and video tamper detection.

Associate Professor, Monash University

PhD · Australian National University

Co-created the FakeBuster deepfake detector and AV-Deepfake1M challenge and dataset.

Associate Professor at Monash University, previously a faculty member at IIT Ropar. He studies affective computing and deepfake detection. He co-created FakeBuster and the AV-Deepfake1M audio-visual deepfake dataset and challenge.

Associate Professor of Computer Science, University of Maryland

PhD · MIT — EECS; co-advised by Muriel Médard and Manolis Kellis

Studies trustworthy AI at UMD. Showed that AI-image watermarks and detectors can be removed or forged. RELAI founder and CSO.

Leads research at the University of Maryland on detecting diffusion-generated images and testing the robustness of watermarking and provenance methods.

Research Scientist, Meta FAIR (Paris)

PhD · Inria Rennes / Meta FAIR (Univ. of Rennes) — Teddy Furon, Hervé Jégou, Matthijs Douze

First author of Stable Signature for images and AudioSeal for localized speech watermarking.

Meta FAIR research scientist who specializes in watermarking for content protection. He is first author of The Stable Signature, which embeds invisible watermarks in latent diffusion models. He also led AudioSeal for localized watermarking of AI-generated speech.

Professor of CS; Director, Maryland Center for Machine Learning, University of Maryland

PhD · UCLA

Developed Tree-Ring Watermarks, invisible and robust fingerprints embedded during diffusion sampling.

Senior author of Tree-Ring Watermarks (NeurIPS 2023). It embeds a Fourier-space pattern in initial diffusion noise. The watermark survives crops, rotations, and color jitter far better than methods applied after generation. He is a leading voice on watermark robustness and model security.

Senior Research Scientist, Apple (AIML)

PhD · University of Seoul

Lead author of RawNet and the AASIST anti-spoofing architecture. Organizes the SASV challenge.

Lead author of the RawNet family and the widely adopted AASIST spectro-temporal graph-attention anti-spoofing model. He organized the first Spoofing-Aware Speaker Verification (SASV) challenge. Previously at Naver Clova and a CMU postdoc, he now works at Apple.

SVP Research, Pindrop

Leads Pindrop’s industrial research in voice anti-spoofing and audio deepfake detection.

Leads Pindrop’s Speech Research team in voice biometrics, deepfake detection, and phoneprinting. He previously researched voice-spoofing countermeasures at Idiap. He is one of the most prominent industrial voices in audio anti-spoofing.

Research Associate, Biometrics Security & Privacy group, Idiap Research Institute

PhD · University of Genoa / National University of Singapore — Computer science (multimedia/video quality) PhD

Idiap researcher in audio-visual deepfake detection and voice and face anti-spoofing. Coauthored an early deepfake benchmark.

Associate Professor, Electrical and Electronic Engineering, Hong Kong Polytechnic University

ASVspoof organizer and coauthor of the t-DCF metric, which accounts for spoofing during evaluation.

A core ASVspoof organizer and t-DCF coauthor. His influential work covers speaker embeddings (i-vector, x-vector, and xi-vector) and speaker verification that accounts for spoofing.

MSU Foundation Professor, Michigan State University

PhD · Carnegie Mellon University

Developed attention-based face-manipulation detection (DFFD) and deep face anti-spoofing.

MSU Foundation Professor and IEEE/IAPR Fellow who leads the Computer Vision Lab. His CVPR 2020 work On the Detection of Digital Face Manipulation introduced an attention-based detector and the DFFD database. It builds on extensive CVPR/ECCV work on face anti-spoofing.

Senior Research Scientist, Head of Biometrics Security & Privacy group, Idiap Research Institute · Professor, School of Criminal Justice, University of Lausanne

PhD · Université de Rennes I (PhD at CNET/France Telecom, 2000) — Signal processing PhD, CNET (France Telecom research center, Rennes)

Heads Biometrics Security & Privacy at Idiap. Studies presentation attacks, morphing, and deepfake detection for biometrics.

Leads biometric anti-spoofing and face presentation-attack detection at Idiap. This work provides a foundation for authentication that resists deepfakes.

Lead Research Scientist; Distinguished Engineer; Head of Creative AI Lab, Sony AI / Sony Group

Senior author of SilentCipher, a deep audio-watermarking method with psychoacoustically imperceptible watermarks.

Leads music and sound research at Sony AI and Sony Group’s Creative AI Lab. He is senior author of SilentCipher (Interspeech 2024). It was the first deep audio-watermarking model to use psychoacoustic-model thresholding for imperceptible, robust watermarks.

Assistant Professor, University of South Florida (Bellini College)

PhD · University of Wisconsin-Madison — Yong Jae Lee

Showed that frozen CLIP features enable fake-image detection to generalize across generative models.

Lead author of Towards Universal Fake Image Detectors (CVPR 2023). It classifies real and fake images in a frozen CLIP feature space. This generalizes far better across unseen GAN and diffusion generators. He earned a PhD at UW-Madison under Yong Jae Lee and joined USF faculty in 2025.

Assistant Professor, Computer Science & Engineering, University of North Texas

PhD · University of Cagliari — Electrical & Computer Engineering (PhD 2010; pattern recognition / adaptive biometrics)

UNT professor who leads the Visual Computing & Biometric Security Lab. Studies fairness and bias in deepfake detection and biometrics.

Senior Researcher / Group Head, Multimedia Security, FAU Erlangen-Nürnberg

PhD · FAU Erlangen-Nürnberg

Heads FAU’s Multimedia Security group. Studies image forensics.

Heads the Multimedia Security group in FAU’s IT Security Infrastructures Lab. He is one of Germany’s leading image-forensics experts. His work detects manipulated images and deepfakes, including through funded deepfake-detection projects.

Full Professor, Institute of Computing, University of Campinas (UNICAMP)

PhD · University of Campinas (UNICAMP)

Founded and heads Recod.ai. Studies digital image and multimedia forensics.

IEEE Fellow and Full Professor of AI and Digital Forensics at UNICAMP, where he heads Recod.ai. He is a long-standing leader in digital image and video forensics. He previously directed UNICAMP’s Institute of Computing.

Assistant Professor, TCG CREST (Institute for Advancing Intelligence)

PhD · Indian Institute of Technology Kharagpur — Dept. of Electronics & Electrical Communication Engineering (speech processing, PhD 2015)

Studies speech anti-spoofing, audio deepfakes, and speaker-verification security. Co-organizes ASVspoof.

Co-organizes ASVspoof and leads work on audio anti-spoofing and speaker verification.

Associate Professor, ECE, Drexel University

PhD · University of Maryland, College Park

Developed the constrained CNN (MISLnet) for universal image-manipulation detection.

Directs Drexel’s Multimedia and Information Security Lab (MISL). His constrained-convolution CNN, MISLnet, is widely used for general-purpose image and video forensics. It was recently extended to identify fingerprints in AI-generated video.

Research Scientist, CSIRO Data61

PhD · Sungkyunkwan University (SKKU) — Simon S. Woo (DASH Lab)

CSIRO Data61 research scientist in deepfake detection and generative-media forensics. Named a 2025 AU AI Rising Star.

Associate Professor, Digital Security Department, EURECOM

PhD · University of Rome Tor Vergata

Coauthored the t-DCF metric and the end-to-end RawNet2 anti-spoofing model.

Co-developed t-DCF and led the application of RawNet2 to end-to-end anti-spoofing from raw waveforms. His work at EURECOM advances deepfake-speech detection and voice-biometric trustworthiness.

Assistant Professor of CS; head of the SPY Lab, ETH Zurich

PhD · Stanford University — Dan Boneh

Studies adversarial security and exposes weaknesses in generative-AI watermarks and detectors.

Leads ETH Zurich’s SPY Lab, which studies adversarial behavior in ML systems. His group published attacks that remove or evade generative-AI watermarks. This work makes him a leading skeptic of watermark robustness.

Full Professor, DEIB, Politecnico di Milano

PhD · Politecnico di Milano

Leads the Image and Sound Processing Group (ISPG). Studies video forensics and deepfake detection.

Full Professor of Telecommunications at Politecnico di Milano who coordinates the Image and Sound Processing Group. He is a long-standing multimedia-forensics researcher in DARPA MediFor/SemaFor. His more than 150 publications cover video forensics and synthetic-media detection.

Project Associate Professor, National Institute of Informatics (Yamagishi Lab)

PhD · SOKENDAI / National Institute of Informatics — Yamagishi Lab

ASVspoof organizer. Pioneered vocoder-generated training data for speech-spoofing countermeasures.

Has served on the ASVspoof organizing team since 2019. His work showed that neural vocoders can efficiently synthesize spoofed training data. This substantially improves the generalization of speech-deepfake countermeasures.

Tenured Associate Professor, Chinese University of Hong Kong, Shenzhen

Directs SCLBD, the lab behind DeepfakeBench and BackdoorBench.

Tenured Associate Professor at CUHK-Shenzhen, previously at Tencent AI Lab. He directs the Secure Computing Lab of Big Data (SCLBD). His group produced DeepfakeBench and BackdoorBench, widely used benchmarks for deepfake detection and backdoor learning.

PhD Candidate, University of Naples Federico II

PhD · University of Naples Federico II — Luisa Verdoliva (GRIP-UNINA)

Lead author of an early study on forensic traces left by diffusion models.

PhD candidate in Luisa Verdoliva’s GRIP lab. Lead author of On the detection of synthetic images generated by diffusion models (ICASSP 2023). It was among the first studies to characterize diffusion-model forensic fingerprints and test detector generalization.

PhD Candidate, University at Buffalo (SUNY)

UB PhD in Lyu’s lab. First author on fairness and generalization in deepfake detection and AI-generated-image detection.

Known for Glaze / DeepfakeBench contributions. Active first author on deepfake-detection generalization papers from Lyu’s lab.

Associate Professor, University of Chinese Academy of Sciences (UCAS)

PhD · Institute of Computing Technology, Chinese Academy of Sciences

UCAS professor and Celeb-DF deepfake-forensics dataset coauthor. Studies video and image coding.

Known for Multi-attentional face forgery detection. Frequently publishes as first author on attention-based detector architectures.

PhD Student, Beijing Jiaotong University

PhD · Beijing Jiaotong University — Yao Zhao

Proposed Neighboring Pixel Relationships (NPR), a universal up-sampling artifact for detection that generalizes across models.

PhD student at Beijing Jiaotong University, advised by Yao Zhao. He is lead author of NPR (CVPR 2024), which models Neighboring Pixel Relationships as a generalized up-sampling artifact. It achieved state-of-the-art detection that generalizes across 28 generative models.

PhD Student, Carnegie Mellon University

PhD · Carnegie Mellon University — Jun-Yan Zhu

Lead author of a CVPR 2020 paper showing that CNN-generated images share a transferable fingerprint.

Lead author of CNN-generated images are surprisingly easy to spot... for now (CVPR 2020). It showed that a detector trained on one CNN generator generalizes to many others. This established the transferable-fingerprint approach. He is advised by Jun-Yan Zhu at CMU.

PhD Student, Nanyang Technological University (S-Lab / MMLab@NTU)

At NTU S-Lab, co-proposed Sequential DeepFake Manipulation detection (Seq-DeepFake / SeqFakeFormer).

Known for Recent CVPR deepfake detection work. Recent first-author work covers robust deepfake-detection benchmarks.

PhD Student, Peking University

PhD · Peking University — Li Yuan, Fanyang Mo

First author of DeepfakeBench (NeurIPS 2023).

PhD student at Peking University, advised by Li Yuan and Fanyang Mo. He is first author of DeepfakeBench (NeurIPS 2023), which became a standard evaluation platform for deepfake detection.