Hong Kong's AI-Generated K-Pop Anti-Drug Ad Backfires and Gets Pulled
A prison service campaign featuring AI-generated K-pop personifications of drugs was withdrawn after viewers found the messaging accidentally glamorized narcotics.
What matters
- Hong Kong's Correctional Services Department produced an AI-generated one-minute anti-drug ad titled "Obsession: The Sugar-Coated Trap."
- The ad featured four K-pop-style characters—Weedy, Icy, Coke, and Little E—personifying cannabis, meth, cocaine, and etomidate.
- Dialogue lines like "If you take me, I'll give you an out of body experience" were widely seen as accidentally glamorizing drugs.
- The prison service pulled the ad after significant online ridicule and backlash.
- The incident illustrates the risks of deploying AI-generated creative content without adequate editorial review.
What happened
Hong Kong's Correctional Services Department released a one-minute AI-generated anti-drug advertisement titled "Obsession: The Sugar-Coated Trap." The video featured four female K-pop-style characters named Weedy, Icy, Coke, and Little E, each personifying a different drug: cannabis, methamphetamine ("ice"), cocaine, and etomidate—a strictly regulated anaesthetic agent used in hospitals that has also been found in vapes.
The intended message was that drugs will ruin lives "no matter how pretty they're packaged." In the ad, the four characters eventually transform into old men behind bars, driving home the consequences of drug use. However, the characters' dialogue—which included lines like "If you take me, I'll give you an out of body experience" (Icy) and "With a romantic puff of smoke, only one stick will help you forget all worries" (Weedy)—struck many viewers as inadvertently glamorizing the very substances the campaign was meant to discourage.
Following significant online reaction and ridicule, the Hong Kong prison service pulled the ad.
Why it matters
The incident highlights a growing risk for government agencies and brands experimenting with AI-generated content: without rigorous creative review and audience testing, AI tools can produce polished, emotionally resonant media that undermines the intended message. In this case, the ad's production quality and K-pop aesthetic may have amplified the appeal of the drug-personifying characters rather than their eventual downfall, creating a tonal mismatch between the campaign's first and second halves.
It also underscores how generative AI can lower the barrier to producing sophisticated video content, but not the editorial judgment needed to deploy it responsibly. Public-sector communicators now have a concrete cautionary tale about what can go wrong when AI-generated creative is rushed to publication without adequate scrutiny of subtext and audience reception.
Public reaction
No Reddit discussion was captured for this story. However, reporting from Channel News Asia indicates the ad drew significant "online reaction and ridicule," and a screengrab circulated on Threads under the account wtfhahameme, suggesting the video became a viral object of mockery before it was pulled.
What to watch
- Whether Hong Kong's Correctional Services Department issues a revised campaign or statement explaining the decision to pull the ad.
- Broader regulatory or internal-policy responses from government agencies regarding the use of generative AI in public messaging.
- Whether other public-sector bodies adjust their AI content review processes in light of this high-profile backfire.
Sources
Public reaction
No Reddit discussion was captured for this story. Reporting indicates the ad generated significant online ridicule, including viral circulation on Threads, before being pulled. The public signal was overwhelmingly mocking rather than supportive.
Signals
- Ridicule and mockery of the ad's unintended glamorization of drugs
- Viral sharing of screengrabs on social platforms like Threads
- Tonal mismatch between the K-pop aesthetic and the anti-drug message
Open questions
- Will the Correctional Services Department release a revised or replacement campaign?
- Did the department conduct audience testing before publishing the ad?
- What AI tools or vendors were used to produce the video?
What to do next
Developers
Review generative AI video outputs for subtext and unintended messaging before integration into public-facing pipelines.
AI-generated creative can carry persuasive subtext that contradicts the intended message, as this anti-drug ad demonstrated.
Founders
Build editorial review and audience-testing checkpoints into any AI content platform targeting government or public-sector clients.
Public-sector AI content failures carry reputational risk; products that bake in review workflows reduce liability.
PMs
Add a 'message alignment' QA step to AI content workflows that flags when generated dialogue or imagery could undercut the campaign objective.
This ad's drug-personifying characters delivered lines that glamorized rather than warned, exposing a gap in creative review.
Investors
Assess whether AI content startups serving regulated industries have adequate guardrails and human-in-the-loop review features.
High-profile backfires like this may drive demand for safer, review-centric AI content tools and create regulatory pressure.
Operators
Pilot AI-generated public messaging with small audience segments and sentiment monitoring before full deployment.
Early sentiment signals could have flagged the ridicule this ad attracted before it reached a wide audience.
Testing notes
Caveats
- This is a news story about a pulled public-service advertisement, not a testable product, API, or developer tool.