How AI Turns Rumours Into “Facts”

Everton’s first match at Hill Dickinson Stadium sparked a headline that sounded wonderfully scientific: 126 decibels. Loudest in the Premier League. Fifth loudest in world football. Precise. Shareable. “Factual”.

Except there was a problem: it was never measured.

In this episode, Dr Nicola Harding uses a seemingly harmless football rumour to unpack something much bigger: how “facts” get manufactured in the age of generative AI. Following the trail from an X account to Grok’s confident answers, Nicola shows how repeated prompts, performative authority, and screenshot culture can turn a claim into a pattern - and then into “truth”.

Drawing on criminology’s concepts of moral panic and deviancy amplification, she maps an emerging loop: data voids get filled by AI, people circulate outputs as evidence, media reports the circulation, and the internet produces new “artefacts” that train the next round of answers. The result isn’t just misinformation - it’s legitimacy laundering.

And when that logic hits fraud and financial crime, the stakes explode. Fraud prevention relies on what we count as true, yet most fraud data is messy, delayed, and probabilistic. When repetition starts masquerading as corroboration, uncertainty gets institutionalised, and criminals get to weaponise “normal” at scale.

This isn’t a panic about AI. It’s a panic produced through AI. And unless we rebuild provenance, slow down decision-making, and treat chatbot outputs as leads (not facts), we’ll keep chasing ghosts that never existed - convinced they were always real.