We examined high-school students’ self-reported deepfake knowledge, recognition ability, attitudes and trust before and immediately after a brief visual and oral information activity. Preserved Qualtrics exports contain 259 pre-test and 253 post-test submissions from October–December 2023. We retain all records as the main descriptive analysis and present October (92/92) and November–December (167/161) results side by side to show variation across collection periods. Selection of the final two knowledge options increased from 12.0% to 59.7%. Selection of the final two recognition-option positions increased from 13.1% to 20.6% overall, but differed by period: 18.5% to 17.4% in October and 10.2% to 22.4% in November–December. No-opinion responses concerning political misuse decreased from 23.6% to 0.8%. We matched students using separately assigned numbers during the study, but the linking list is unavailable for reanalysis. Findings concern immediate self-reports; the design has neither an untreated comparison group nor an objective detection task.
Download Full Paper (PDF)