benchmark arXiv Mar 2, 2026 · 5w ago
Michael Rettinger, Ben Beaumont, Nhien-An Le-Khac et al. · University College Dublin
Benchmarks six public deepfake detection tools with law enforcement investigators, finding humans outperform all automated AI classifiers and forensic tools
Output Integrity Attack visiongenerative
The proliferation of deepfake imagery poses escalating challenges for practitioners tasked with verifying digital media authenticity. While detection algorithm research is abundant, empirical evaluations of publicly accessible tools that practitioners actually use remain scarce. This paper presents the first cross-paradigm evaluation of six tools, spanning two complementary detection approaches: forensic analysis tools (InVID \& WeVerify, FotoForensics, Forensically) and AI-based classifiers (DecopyAI, FaceOnLive, Bitmind). Both tool categories were evaluated by professional investigators with law enforcement experience using blinded protocols across datasets comprising authentic, tampered, and AI-generated images sourced from DF40, CelebDF, and CASIA-v2. We report three principal findings: forensic tools exhibit high recall but poor specificity, while AI classifiers demonstrate the inverse pattern; human evaluators substantially outperform all automated tools; and human-AI disagreement is asymmetric, with human judgment prevailing in the vast majority of discordant cases. We discuss implications for practitioner workflows and identify critical gaps in current detection capabilities.
cnn transformer University College Dublin
ML Security Papers