A government-backed AI-driven anti-drug campaign has inadvertently glamourised illegal substances, according to a damning report from UK digital ethics experts. The campaign, which used generative AI to create hyper-realistic videos depicting the euphoric highs of drug use before transitioning to negative consequences, has been found to actually increase curiosity among young viewers.
The initiative, part of a broader push to leverage artificial intelligence for public health messaging, employed large language models to script narratives and deepfake technology to produce compelling visuals. However, a study by the Oxford Internet Institute reveals that the 'shock factor' approach backfired: the initial pleasurable scenes were too enticing, and the subsequent warnings failed to deter. 'The brain's reward system is faster than its risk assessment,' explains Dr. Helena Marsh, lead researcher. 'The AI-generated content became a free advertisement for the very behaviours it aimed to prevent.'
The campaign's failure underscores a broader challenge in AI ethics: the uncanny valley of persuasion. When AI crafts messages that are too perfect, they lose the authenticity needed for credible warnings. Furthermore, the algorithm's optimisation for engagement inadvertently prioritised the most shareable, and thus sensational, clips. 'We built a machine that amplifies desire, not reflection,' notes Julian Vane, a technology and innovation consultant. 'This is a textbook case of unintended consequences in algorithmic design.'
The report calls for a moratorium on AI-generated public health campaigns until robust ethical frameworks are in place. It recommends human oversight of creative decisions, and a shift from fear-based messaging to positive reinforcement strategies. 'You cannot fight fire with jet fuel,' Vane adds. 'AI is a tool for precision, not brute force propaganda.'
The government has paused the campaign pending review, but the damage to public trust may be lasting. As AI becomes more embedded in our lives, this incident serves as a cautionary tale: technology that mimics human emotion can easily become a mirror for our worst impulses.
