The Hong Kong government removed an AI-generated anti-drug advertisement after internet users mocked the video for making narcotics look appealing [1].
The incident highlights the risks of using generative AI for public health messaging, where a lack of human nuance can lead to the opposite of the intended effect. By attempting to appeal to youth through a high-gloss aesthetic, the campaign inadvertently glamorized the behavior it sought to prevent.
The Census and Statistics Department released the video, which featured a glittery AI-generated pop-girl group, to warn young people about the dangers of drugs [1]. The campaign was timed to coincide with the International Day Against Drug Abuse and International Trafficking, which occurs on June 26 [2].
Public reaction was swift and critical. Some viewers described the video as "the best drug ad ever" [3], while others noted that the visual style resembled a corporate beverage commercial. One social-media commentator said the ad looks more like a Coca-Cola commercial than a warning about drugs [1].
Due to the backlash, the government pulled the video on June 29, 2024 [3]. While some reports suggest the ad debuted on June 26, other sources state it was released on Friday, June 28, 2024 [1].
This failure is not an isolated event for the region's public health outreach. Tom Grundy, Editor-in-Chief of Hong Kong Free Press, said it is not the first time an anti-drug campaign has gone wrong in Hong Kong [4].
“The ad looks more like a Coca-Cola commercial than a warning about drugs.”
The failure of this campaign underscores a growing tension in government communications as they adopt generative AI to reach younger demographics. When AI optimizes for 'engagement' or 'aesthetic appeal' without a deep understanding of social stigma and irony, it can create a 'boomerang effect' where the target audience interprets the message as an endorsement rather than a warning.
