Researchers analysed 112 pieces of fake content created or altered by AI and circulated in the UK between January 2025 and March 2026, and identified eight areas of harm, ranging from riots to scams.
Peter Cunliffe-Jones, Joseph O'Leary, David Corney and Andrew Dudfield have collected 112 examples of AI-generated or modified disinformation or misinformation, which circulated in the UK between 1 January 2025 and 31 March 2026 and accumulated tens of millions of views. The authors caution that this is not an exhaustive inventory, but a snapshot intended to illustrate the range of possible effects.
Eight areas of damage
The analysis highlights a real risk of contributing to harm in eight distinct areas: serious disturbances to public order and self-defence violence, direct damage to health, financial losses related to online scams, harassment violent enough to affect the health or behaviour of the people targeted, disruption of the public's relations with the police and the judiciary, adherence to dangerous conspiracy theories, and slower shifts in social and political opinions.
More than four out of five pieces of content also fuel a general mistrust: they are not simply imprecise, but frankly false or misleading, which gives the public good reason to be wary of information in general. The authors propose a ranking grid for these risks, useful to regulators and platforms.
Read the full article on the original source
Go to the source →