JudgeGPT Human Perception Data: Dual-Axis Judgments of AI-Generated Disinformation

Large-scale human perception data on AI-generated vs. human-written text, collected through the JudgeGPT socio-technical evaluation platform. 539 participants assessed text fragments on two continuous axes, origin (human vs. machine) and veracity (legitimate vs. fake), producing 2,546 judgments while demographic, behavioral and temporal data were recorded. Includes the RogueGPT stimulus corpus (3,278 multilingual fragments). Snapshot of 11 March 2026. This version supersedes the initial release of 504 participants and 2,438 judgments. Two properties matter for reuse: the stimulus pool is 98 percent machine-authored, so pooled origin accuracy must not be read against a 50 percent baseline, and 117 participants aged 16 to 18 stem from a single school-class recruitment event and are not independent draws. See the codebook for both. Created as part of a doctoral dissertation on the AI-driven disinformation ecosystem at Frankfurt University of Applied Sciences.