BACKGROUND: Machine learning and AI-enabled tools are emerging as powerful resources for understanding online tobacco marketing. This study aimed to characterize the types of advertisements found in Indonesian digital newspaper platforms and identify the targeted audience or user profiles.
METHODS: We developed specialized scraper engines designed around five fictional user profiles: two females and three males. These profiles simulated real-world digital behaviors across platforms like Google and e-commerce sites. The engine, powered by advanced AI technology and Optical Character Recognition (OCR), automatically classified tobacco advertisements in real time. News articles in both English and Bahasa from 22 newspapers with the highest digital reach were analyzed. Reach estimates were based on the number of monthly visitors to each newspaper.
RESULTS: From March to May 2023, 42.418 English and Bahasa articles were reviewed, and 306.874 ads were scanned, identifying 5.502 tobacco advertisements. Detik Online, with over 43 million readers, hosted the most ads, primarily in its "Letters from Readers" section, featuring PT HM Sampoerna’s cigarette brands. A 20-year-old female profile experienced nearly five times more tobacco ads (24%) than a 23-year-old male profile (5%), with most ads in video format on smartphones. Tobacco ads primarily targeted males (60%), but all profiles had highest exposure to Sampoerna A Mild (42%) and IQOS (19%) ads.
CONCLUSIONS: By leveraging the credibility of traditional print media and the vast reach of the internet, tobacco advertisers expose larger audiences to their products. The transient and elusive nature of digital ads complicates documentation and enforcement of restrictions on tobacco advertising, promotion, and sponsorship (TAPS). Further research and robust digital monitoring systems are essential to strengthen enforcement. Targeted advertising towards young women poses significant public health risks, potentially exacerbating Indonesia's high smoking rates.