Preventing fraud in a post-COVID era of online HIV research: two case studies and their lessons for future research

Internet-based research methods have expanded researchers' ability to reach marginalized populations, including sexual and gender minority (SGM) individuals, for HIV prevention studies. However, the post-COVID proliferation of online research has heightened the vulnerability of these studies to fraudulent and duplicate participation. Despite growing documentation of this problem, practical guidance from large-scale HIV cohort studies remains limited. We present two case studies from NIH-funded, limited-interaction cohort studies—the American Transformative HIV Study (AMETHST) and Keeping it LITE − 2—both of which recruited thousands of SGM individuals via geosocial and social media platforms, conducted remote surveys, and utilized mailed at-home specimen collection. We describe the fraud-prevention strategies each study deployed, the methods by which fraudulent participants were detected, and the protocol adaptations made in response. Both studies implemented layered fraud-prevention approaches combining automated safeguards (CAPTCHA, IP tracking, duplicate detection, unique survey links) and systematic manual review. Despite these measures, AMETHST identified 177 fraudulent participants—the majority traceable to a single publicly accessible static recruitment link—plus another 305 duplicates. Keeping it LITE − 2 identified and excluded 483 individuals for fraudulent or duplicate activity. Fraudulent tactics included fabricated addresses, recycled HIV test images, falsified identity documents, and, notably, AI-generated prescription photographs. No single fraud-prevention strategy is sufficient. Effective protection requires defense-in-depth, combining automated systems with ongoing manual review and rapid-response capacity. The emerging use of generative AI to fabricate verification documents represents a growing threat that the field must proactively address.