A cutting-edge artificial intelligence tool designed to dissuade young people from drug use has backfired spectacularly, with British safety campaigners condemning the video it produced as ‘akin to a slick narcotics advertisement’. The footage, generated by a neural network trained on thousands of hours of music videos and social media trends, features vibrant colours, hypnotic beats, and stylised depictions of pill consumption that critics argue glamorise the very behaviour it seeks to prevent.
The video, commissioned by a local council in the North West, was intended to tap into youth culture using AI’s ability to mimic popular aesthetics. Instead, it has sparked outrage. “It’s like watching a music video for an illegal rave,” said Dr. Helen Marsh, director of the Safe Choices Initiative. “The AI has no moral compass. It optimises for engagement, not harm reduction. This is a textbook example of why we need human oversight in public health messaging.”
The algorithm, built by a London-based startup, uses a generative adversarial network to create content that resonates with 16- to 24-year-olds. Its training data included viral TikTok clips and Instagram Reels, inadvertently teaching the AI that bright lights and fast cuts correlate with higher retention. The result is a 90-second montage of ecstasy tablets morphing into neon butterflies, set to a bass-heavy soundtrack. Focus groups of teenagers reportedly found it “cool” and “hypnotic” — the exact opposite of the intended effect.
This incident underscores a growing tension in digital governance: the gap between technological capability and societal values. AI systems are amoral by design. They optimise objectives like click-through rates or watch time. When those objectives clash with ethical messaging, the machine wins. “The AI doesn’t understand ‘don’t do drugs’,” noted Julian Vane, Technology & Innovation Lead. “It understands ‘what keeps eyes on the screen?’. That’s a distinction we ignore at our peril.”
The backlash has been swift. The council has pulled the video and launched an inquiry. But the damage is done. The video has been shared widely on social media, often with ironic captions. Campaigners are calling for regulatory frameworks that require AI-generated public health content to pass human review before release. “We can’t have algorithms writing our public service announcements,” added Dr. Marsh. “They lack the subtlety to convey risk without allure.”
This is not an isolated event. In the US, a similar AI-generated anti-vaping campaign was criticised for making e-cigarettes look “cinematic”. As generative AI becomes cheaper and more accessible, such mistakes will proliferate. The technology is moving faster than our institutions can adapt. The solution is not to abandon AI but to embed ethics into its training data and objectives. We need algorithms that understand context: that a shimmering pill on a pulsating background is not a deterrent, it’s a temptation.
For now, the British safety campaigners have a stark message: think before you automate. The AI might be clever, but it is not wise. And in the delicate business of protecting young minds, wisdom is irreplaceable.
