Unboxing unhealthy food marketing: An analysis of the extent and nature of unhealthy food marketing to children on YouTube in Canada using the CHOPPED artificial intelligence system

Background: Unhealthy food marketing negatively affects children's dietary patterns. However, little is known about the extent and nature of unhealthy food marketing to children on YouTube due to reliance on manual assessment methods. Objectives: To use a novel artificial intelligence (AI) system (CHild Oriented food Product Promotion Evaluation and Detection system, CHOPPED) to examine the extent and nature of unhealthy food marketing in YouTube videos suggested to Avatar accounts created for children and adults based on age and gender. Methods: Over two months of continuous monitoring, CHOPPED collected marketing instances played before, during, and after videos suggested by YouTube's algorithm to six child and two adult accounts that were not signed in and differed by age (2-4, 5-8, 9-12, =<18 years) and gender across four Canadian cities (Calgary, Vancouver, Toronto, Halifax). CHOPPED characterized the extent and nature of monthly child-targeted unhealthy food marketing instances during 24-hour and typical 2-hour daily viewing periods. Results: CHOPPED collected 60,633 marketing instances. No food marketing was shown to child accounts 2-4 years. On average, child accounts (5-12 years) were shown 25 monthly unhealthy food marketing instances during a 2-hour daily viewing period, 24 of which were child-targeted. The boy's account (9-12 years) was shown a significantly higher monthly frequency of child-targeted unhealthy food marketing instances (340, 70.2%) than the boy's account (5-8 years; 241, 66.2%), the girl's account (5-8 years; 261, 60.6%), the man's account (224, 57.7%), and the woman's account (263, 54.7%) during a 24-hour (p<0.05) but not a 2-hour daily viewing period. Monthly frequencies of some core child-targeted techniques shown to the boy's account (9-12 years) were also higher than other accounts during a 24-hour (p<0.05) but not a 2-hour daily viewing period. Conclusions: This study pioneered AI-driven analysis of the extent and nature of unhealthy food marketing to children on YouTube. Children's potential exposure to such marketing on YouTube may be low. There was some evidence that the older boys' account was shown more monthly child-targeted unhealthy food marketing instances and of some core child-targeted techniques than other accounts. On YouTube, food marketing may target a general audience rather than children by age and gender.

Authors

Institutions

Publication Details

Journal
medRxiv
Published
2026-11-15
DOI
https://doi.org/10.64898/2026.11.12.26367461
Primary Topic
Consumer Attitudes and Food Labeling
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Unboxing unhealthy food marketing: An analysis of the extent and nature of unhealthy food marketing to children on YouTube in Canada using the CHOPPED artificial intelligence system

Shelly Yin, Charlene D Elliott, Dana Lee Olstad, Zahra Shakeri Hossein Abad et al.
medRxiv
Consumer Attitudes and Food Labeling
preprint

Unboxing unhealthy food marketing: An analysis of the extent and nature of unhealthy food marketing to children on YouTube in Canada using the CHOPPED artificial intelligence system

Shelly Yin, Charlene D Elliott, Dana Lee Olstad, Zahra Shakeri Hossein Abad, Monique Potvin Kent, Tyler S. Williamson, Rachel J. L. Prowse, Leticia Radin Pereira, Sara Al Sharani, Sadeeb Hossain, Spencer Fietz, Inara Lalani
preprint en

Abstract

Background: Unhealthy food marketing negatively affects children's dietary patterns. However, little is known about the extent and nature of unhealthy food marketing to children on YouTube due to reliance on manual assessment methods. Objectives: To use a novel artificial intelligence (AI) system (CHild Oriented food Product Promotion Evaluation and Detection system, CHOPPED) to examine the extent and nature of unhealthy food marketing in YouTube videos suggested to Avatar accounts created for children and adults based on age and gender. Methods: Over two months of continuous monitoring, CHOPPED collected marketing instances played before, during, and after videos suggested by YouTube's algorithm to six child and two adult accounts that were not signed in and differed by age (2-4, 5-8, 9-12, =<18 years) and gender across four Canadian cities (Calgary, Vancouver, Toronto, Halifax). CHOPPED characterized the extent and nature of monthly child-targeted unhealthy food marketing instances during 24-hour and typical 2-hour daily viewing periods. Results: CHOPPED collected 60,633 marketing instances. No food marketing was shown to child accounts 2-4 years. On average, child accounts (5-12 years) were shown 25 monthly unhealthy food marketing instances during a 2-hour daily viewing period, 24 of which were child-targeted. The boy's account (9-12 years) was shown a significantly higher monthly frequency of child-targeted unhealthy food marketing instances (340, 70.2%) than the boy's account (5-8 years; 241, 66.2%), the girl's account (5-8 years; 261, 60.6%), the man's account (224, 57.7%), and the woman's account (263, 54.7%) during a 24-hour (p<0.05) but not a 2-hour daily viewing period. Monthly frequencies of some core child-targeted techniques shown to the boy's account (9-12 years) were also higher than other accounts during a 24-hour (p<0.05) but not a 2-hour daily viewing period. Conclusions: This study pioneered AI-driven analysis of the extent and nature of unhealthy food marketing to children on YouTube. Children's potential exposure to such marketing on YouTube may be low. There was some evidence that the older boys' account was shown more monthly child-targeted unhealthy food marketing instances and of some core child-targeted techniques than other accounts. On YouTube, food marketing may target a general audience rather than children by age and gender.

medRxiv
Memorial University of Newfoundland (CA), University of Calgary (CA), University of Toronto (CA)
Consumer Attitudes and Food Labeling
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon