(Artificial intelligence) AI-generated sexual videos made without the permission of the people depicted increased from an estimated 500,000 in 2023 to 8 million in 2025, according to estimates cited in a Lancaster University research release. Most of the content depicts women and girls.
These videos are known as deepfakes. They use AI to make someone appear in a video they never took part in. For example, AI can put a person’s face onto someone else’s body in a pornographic video, making it look as though that person was involved.
“Anyone with a single photo online can be quickly and easily targeted in content that falsely depicts them in sexual or intimate contexts, of any nature, without their consent. Emerging evidence demonstrates the wide-ranging and persistent harms of this digital abuse,” said Sarah Morgan from Lancaster University.
Morgan led the study, “The Backbone of Abuse,” with Dr. Sophie Nightingale, also from Lancaster, and Professor Hany Farid of Dartmouth College. It was published in the Journal of Online Trust and Safety.
The study looked at the technology services that keep websites carrying this material online. The figures of 500,000 and 8 million came from estimates cited in the release, rather than a video count conducted by the research team.
Researchers identified 88 websites carrying AI-generated sexual images or videos made without consent. Of these, 38 carried only this type of content. Some websites had hundreds of thousands of videos.
Nearly all the material featured women. Those depicted included Hollywood and Hindi-film stars, K-pop idols, feminists, and activists.
The team identified five major technology providers whose services helped these websites operate. Cloudflare was the name that appeared most often. The research covered providers supporting the sites either knowingly or unknowingly; it did not establish that they deliberately helped spread the abuse.
Websites depend on several services to stay online. Hosting services store their files. Domain name servers help a browser find the right website when someone enters its address. Content delivery networks help send images and videos to viewers.
The researchers also found that 93% of the websites could be found through Google Search, making the abusive material easy to discover.
Morgan urged technology providers to block websites devoted entirely to fake sexual content made without consent. For websites that also carry other material, she recommended suspending services until their operators make good-faith efforts to remove the abuse.
The researchers also proposed using a tool called “perceptual hashing.” It gives an image or video a digital fingerprint that systems can recognize. Once material is confirmed as abusive, that fingerprint can help detect matching or similar copies and prevent them from being shared again.
Morgan said laws could discourage some offenders, but providers also needed to help prevent the material from spreading.
Nightingale said people depicted in these videos should not be blamed.
“All too often the victims of abuse that is sexual in nature are attributed responsibility — there is an urgent need to change this. Blame should be assigned to the perpetrators who create, distribute, and facilitate the sharing of fake content that has potential to ruin lives.”