Ofcom has launched an urgent investigation after a government-funded AI-generated anti-drug video spectacularly backfired, reportedly making narcotics look appealing to viewers. The short film, produced by a London-based agency using generative AI tools, was intended to deter young people from drug use but instead garnered criticism for its glamorised depiction of substance abuse.
The video, which uses hyper-realistic AI to create vivid scenes of drug-induced euphoria, has been described by experts as a 'textbook failure' of algorithm-driven messaging. Critics argue that the AI's optimisation for engagement inadvertently prioritised aesthetic appeal over deterrent messaging. 'The AI was trained on thousands of hours of cinematic content, where drug use often appears glamorous,' explained Dr. Mira Patel, a digital ethicist at King's College London. 'It inadvertently reproduced those tropes, mistaking viewer retention for effective messaging.'
The video, which has since been taken down, featured slow-motion sequences of colourful pills and shimmering powders, accompanied by a pulsating soundtrack. Focus groups reported feeling 'curious' about the substances depicted, rather than repulsed. One teenager quoted in the report said, 'It looked like a music video. I wanted to try whatever they were taking.'
Ofcom's investigation will focus on whether the content violated broadcasting standards regarding harmful material. The regulator has powers to fine producers and mandate corrective messaging. 'We are deeply concerned that an anti-drug campaign could have the opposite effect,' said an Ofcom spokesperson. 'We will be examining the algorithms and human oversight involved.'
This incident highlights a growing concern around AI-generated content in public service messaging. Advocates for digital sovereignty argue that such tools should be subject to rigorous testing before deployment. 'We cannot outsource our values to code,' said Julian Vane, a former Silicon Valley technologist now advising on ethical AI. 'When we let algorithms optimise for metrics like watch time, we lose sight of human context. A machine doesn't understand that making drugs look cool is dangerous.'
The failure also raises questions about the procurement process. The agency used a generative AI model fine-tuned on Hollywood productions, which researchers say contains inherent biases towards dramatic visualisation. 'This is a classic case of garbage in, garbage out,' added Vane. 'If you train an AI on Scarface and Trainspotting without filtering, don't be surprised when it makes heroin seem edgy.'
The government has distanced itself from the creative decisions, stating that the video was produced before the recent introduction of binding AI ethics guidelines. However, the opposition has called for a full parliamentary inquiry. 'This isn't just a waste of public funds; it's a public health risk,' said Shadow Digital Secretary, Maya Singh. 'We need a moratorium on AI-generated public service announcements until proper safeguards are in place.'
Meanwhile, tech commentators are using the debacle as a case study for AI's limitations in nuanced social messaging. 'The user experience of society is at stake here,' Vane concluded. 'We need to design AI that explains the consequences of drug use, not one that unwittingly markets them.'
Ofcom's report is expected within three months. In the meantime, the video has become a viral meme, ironically beloved by the very demographic it was meant to protect.
