Study reveals widespread abuse involving non-consensual AI-generated intimate images

Around one in 15 adults has experienced AI-generated image-based abuse, according to a new study.

The general-population survey of more than 7,200 adults in the United States, the United Kingdom and Australia found 7% had experienced at least one form of AI-generated image-based abuse since turning 18.

This included having an intimate image of themselves created or altered using AI without their consent, having an AI-generated intimate image shared without their consent, or being threatened with the sharing of such an image.

Almost 87% of victim-survivors reported at least one impact on their lives, including effects on their health and wellbeing, relationships, work or finances.

Lead author Professor Nicola Henry from Australia's RMIT University conducted the research with support from Google, and said the findings revealed the extent of the abuse across the wider community.

"We now have evidence of how widely AI-generated image-based abuse is affecting people across the community, not just celebrities and public figures," she said.

"The true scale may be even greater because some people may never know that AI-generated images of them have been created or shared."

Higher rates of AI-generated abuse were reported by people under 35, people from racial and ethnic minority backgrounds, men, and people with disability.

Men reported experiencing the abuse at more than twice the rate of women, at 10.2% compared with 3.8%.

However, women reported more negative reactions and impacts, including fear for their safety and effects on their mental and physical health. This may reflect broader gender inequalities, stigma and safety risks that women disproportionately face.

Henry said the findings show that anyone can be affected by this abuse, but both the risk of experiencing it and its consequences are not evenly distributed.

"As generative AI becomes more widely available, there is greater potential for these tools to be misused to create, share or threaten to share intimate images without consent," she said.

"What we don't yet know is how quickly that is translating into increased victimization."

The latest findings, published in the International Journal of Human–Computer Interaction, build on two earlier studies using the same international dataset.

One earlier study found that 3.2% of respondents reported creating, sharing or threatening to share AI-generated sexual images without consent.

Among those who shared the material, the most commonly reported motivation was "for fun or as a joke."

Another study found that almost one in five adults had deliberately viewed AI-generated sexual images of real people, excluding those who did so solely to report or remove the material.

Henry said the findings strengthened the case for greater responsibility on technology companies to prevent AI-generated image-based abuse before it occurs.

"Australia's proposed digital duty of care is an opportunity to shift the focus from responding to abuse after it happens to requiring online services to prevent foreseeable harm," she said.

"But regulation and new laws are not enough. We also need better education, accessible support for victim-survivors, and greater responsibility on technology companies and developers to build safety into AI tools and apps from the outset."

 "We've worked with Google researchers for a long time to better understand this broader issue and identify where industry-wide prevention efforts are most needed."

Co-author Dr Rebecca Umbach, Staff User Experience Researcher at Google, said the findings provided important new evidence about this form of abuse.

"Understanding who is experiencing AI-generated image-based abuse, how it occurs and the impacts it has on people is essential if we are going to design more effective prevention and responses," Umbach said.

"Research that brings together academic and technology-sector expertise can help identify where interventions are most needed and how technology itself can be part of preventing harm."

The international team is continuing its work to better understand and prevent AI-generated image-based abuse. The research program was funded by Google.