Survey: 7% of Adults in US, UK and Australia Hit by AI Image Abuse
Abuse built on AI-generated images reaches well beyond celebrities. A peer-reviewed survey in three countries suggests about one adult in 15 has been a target, and the authors think the true figure could be higher.
The headline number
Researchers questioned more than 7,200 adults in the United States, the United Kingdom and Australia. Seven percent said they had been affected at least once since they turned 18.
The study counts three kinds of experience: an intimate picture of the person made or altered with AI without consent, one that was then shared, and a threat to share one.
Lead author Nicola Henry, a professor at RMIT University in Australia, said some victims may never find out that such images of them exist. That is why she believes the real scale may be greater.
Men more often, women harder hit
The split by gender runs two ways. Men reported being affected at 10.2%, against 3.8% for women, so more than twice the rate. Under-35s, people from racial and ethnic minorities and people with disabilities also reported higher rates.
Women who were affected, though, described stronger negative reactions. They spoke of fear for their safety and harm to both their mental and physical health. The study says this could stem from wider gender inequality, stigma and the risks women disproportionately face.
Across all victim-survivors, almost 87% named at least one effect on their lives. The list covers health and well-being, relationships, work and money.
What Henry calls for
Henry argued that new laws will not be enough by themselves. Her list also includes education, accessible support for victim-survivors, and tech firms building safety into their AI tools from the start. She sees Australia's proposed digital duty of care as a way to move from reacting to harm towards requiring services to prevent foreseeable harm.
On how fast the problem is spreading, she was careful: "What we don't yet know is how quickly that is translating into increased victimization."
The research was done with support from Google. Co-author Dr Rebecca Umbach works at Google as a staff user experience researcher. She called the findings important new evidence.
Two earlier papers from the same data
The new results appear in the International Journal of Human–Computer Interaction. Two earlier studies used the same international data set.
One looked at perpetration. It found that 3.2% of respondents had created, shared or threatened to share AI-generated sexual images without consent. Among those who shared, the most common reason given was "for fun or as a joke".
The second dealt with viewing. Almost one in five adults had knowingly viewed AI-made sexual images of real people, leaving out anyone who looked only in order to report or remove them.
Our view
The survey asks people about what happened to them. Henry herself notes that it is not yet known how quickly these experiences are growing. Google's support and a Google employee among the co-authors are part of the context for how the work is read.
Source: Phys.org