TL;DR

A survey in the International Journal of Human-Computer Interaction, reported on 3 November, finds 7% of adults in the UK, US and Australia have experienced AI-generated image abuse as adults. RMIT University’s Professor Nicola Henry led the work, and Google funded it. The paper shows two things the university’s release does not: the fieldwork dates from July and August 2024, and the odds of this abuse were higher for people in the UK than in the US.

The figure, reported by News-Medical from RMIT’s release, covers three experiences: an intimate image made or altered with AI without consent, such an image being shared, or a threat to share one. Across 7,230 respondents, on weighted data, about 4.3% said someone had created such an image of them, 4.4% said one had been shared and 3.6% reported a threat. Almost 87% of those affected described at least one effect on health, relationships, work or money.

A result that cuts against assumptions

Men were far likelier than women to report this abuse, at 10.2% against 3.8%. Women reported more severe reactions, such as safety fears. Henry says anyone can be affected, but that neither the risk nor the consequences are evenly spread. Rates were also higher among under-35s, disabled people and ethnic minority respondents.

What it says about the UK

The paper offers a UK signal that the release leaves out. On a broader measure covering any digitally altered or fake intimate image, AI or not, the UK had the highest rate of the three countries at 17.2%, against 14.5% in the US. In the regression analysis, Americans had lower odds of AI-related victimisation than people in either the UK or Australia. On the 7% measure itself, the paper’s regional table puts the UK at 9.3%, against 7.7% in Australia and 6.5% in the US.

Reading it carefully

Two cautions apply. The fieldwork is more than two years old, so it may not reflect today’s tools, and Henry notes that some people never learn such images of them exist. And Google’s role is substantial: an unrestricted gift from the company paid for the programme, and a Google researcher is a co-author.

Henry’s policy hook is Australia’s planned duty of care for digital services. In our view, UK readers can ask the equivalent question of their own regime, as the Suffolk councillor case has done from a different direction. For employers, the finding that this reaches ordinary people, and not only public figures, is a reason to cover it explicitly in conduct and support policies.