Recent Meta platform Despite Meta’s policy banning sexually explicit advertising, the company allowed an advertisement for an AI pornography generator to run on its platform. The ad appeared to encourage users to create deepfake videos featuring women who resemble prominent female US politicians. The incident is the latest in a series of reported failures involving advertisements for tools that generate non-consensual intimate images.
The tool, called Kromix, describes itself as an “AI image styler.” In one video ad reviewed by WIRED, a voiceover promotes the service as having “no limits,” says that “all the characters are real people,” and calls it “real AI used by men.” The ad shows a woman resembling a prominent American politician standing in front of a US flag. After the caption “What if she moves?” appears, the woman winks before the video cuts to an explicit scene featuring a person with the same face.
The Kromix app, which was available on Apple’s App Store before WIRED requested comment, encouraged paying users to upload photos for use in scenarios with names such as “Bedroom Rape” and “Disney Love.” The app’s promotional videos also included an AI-generated pornographic video showing a woman wearing a Spider-Man costume. Contact information listed for the app did not respond to requests for comment.
Apple has faced repeated challenges removing similar AI applications from its App Store. Critics say some of these tools violate the company’s policies on non-consensual intimate imagery, nudity and sexually explicit content. This summer, San Francisco’s city attorney sent Apple a cease-and-desist letter demanding the removal of 13 specific apps and accusing the company of “aiding and abetting” the sale of explicit deepfake images.
Meta’s ad library showed that the Kromix advertisement targeted male users with the tagline “Unleashing the magic of next-generation AI.” The ad was removed after WIRED contacted Meta. The company said accounts linked to the campaign had run 32 advertisements, with individual ads active for between five and 46 hours. Most of the ads received 10 or fewer impressions.
The advertisement was identified by Katie Paul, director of the Tech Transparency Project. “All of the ads we’ve seen in connection with the nudify issue exclusively sexualize, strip and subject women to AI-generated pornography,” Paul said. “This raises serious questions about how Meta intends to address this content—not only generally, but also when it targets female politicians at a time when Meta is actively allowing these advertisements to appear on its platforms.”
Politicians and elected officials—many of them women—have been targeted by non-consensual deepfake images and videos for years. In January, authorities accused Grok of being used to create approximately 3 million sexually explicit images of women and girls. Swedish, English and Northern Irish officials have also been targeted by people using the technology, often with official photographs serving as source material.
“As the creation and distribution of AI-generated content has been democratized, previously innocuous media such as official portraits and press conference videos have become the raw material for low-code and no-code tools that can be used by anyone with an internet connection,” said Benjamin Schultz, principal investigator at the American Sunlight Project. Schultz is examining how harmful deepfake technologies are being used to target politicians and public figures.
According to Meta’s public information, its advertising standards require every ad to undergo review before appearing on platforms such as Facebook and Instagram. Meta says its system relies primarily on automated tools to check advertisements and business assets against its policies, with reviews beginning before an ad starts running. The page that hosted the Kromix campaign was created on August 3, shortly before the ad began running, and had no followers—details that could have indicated the account was suspicious.
Source: www.wired.com