Diamond-Smith et al. evaluate the effectiveness of using social media as a recruitment method for a COVID-19 and social networks study pertaining to Indigenous and Native American communities in California. Diamond-Smith et al. recruited participants through Facebook, Instagram ads, community partner outreach, and a Qualtrics research panel. They collected data using a 30 minute online survey with 50 questions, and identified fraudulent responses using re-CAPTCHA, IP geolocation, and other verification methods. In total, 5099 survey responses were received, however, 3926 were excluded from the data due to fraudulence or poor quality. As a result, the final data consisted of 137 Native Americans and 72 Mesoamerican Indigenous participants. Overall, it was seen that recruiting Native American and Mesoamerican Indigenous communities through social media was ineffective. Even when using screening tools such as CAPTCHA and honeypot questions, a large proportion of the responses were either fraudulent or generated by AI bots. These bots were successful at bypassing most security methods, which resulted in many data entries being rechecked or discarded completely. Ultimately, Facebook was an ineffective recruitment method, whereas the research panel yielded the most authentic data at a lower cost, although the number of eligible participants remained limited. The authors conclude that not only is it challenging to recruit hard-to-reach populations through the use of social media, but AI bots are becoming sophisticated enough to interfere with recruitment efforts, making it harder to collect authentic data from real people. New outreach strategies and stronger verification tools are needed in order to maintain data quality in online public health research.