- Nudify app ads on Facebook and Instagram promoted AI tools that appeared to sexualize female US politicians without consent.
- Meta removed the nudify app ads after questions, yet its own policies already ban sexual imagery and nudification promotions.
- The campaign used innocent-looking video thumbnails to conceal pornographic material, a familiar tactic for evading automated ad reviews.
- Apple also faces mounting pressure to prevent App Store software from producing nonconsensual intimate images and sexual deepfakes.
Table of Contents
Nudify app ads got past the guardrails
Meta has spent years telling lawmakers, advertisers, and the rest of us that its automated systems can police harmful content at enormous scale. Then nudify app ads appear on Facebook and Instagram promoting a tool that seemingly lets users turn women, including female US politicians, into pornography. That is not a marginal failure. It is a direct hit on rules Meta says it already enforces.
The ads promoted Kromix, an app marketed as an AI image-editing tool. But the pitch reportedly went much further: one video described an AI with “no restrictions,” claimed its characters were real people, and called it “the AI that men actually use.” Another showed a woman resembling a prominent American politician before cutting to a pornographic scene using the same apparent likeness.
The campaign was reportedly aimed only at male users. That detail matters. These weren’t generic ads for a questionable photo filter that happened to be misused; the product messaging appears built around the sexual exploitation of identifiable women. These nudify app ads were designed to market that exploitation directly. When the targets include public officials, the damage spills beyond privacy into political intimidation. A politician who knows any official photo can become raw material for fake porn has one more ugly reason to avoid public life.
Meta removed the ads after WIRED sought comment. According to the company, the associated account ran 32 ads for between five and 46 hours, and most drew fewer than 10 impressions. Small reach is better than viral reach, obviously. But it is not much of a defense when a platform’s review machinery approved the campaign in the first place. Even limited distribution of nudify app ads can validate and finance abusive products.
The problem with automated ad review
Meta’s advertising documentation says ads are reviewed before they run, primarily through automated tools. Its policies prohibit sexual imagery in ads, while its adult sexual exploitation rules explicitly bar nonconsensual intimate imagery and promotion of nudify services. The rules leave little room for confusion. Yet in practice, nudify app ads keep finding the gaps.
Part of the trick is visual misdirection. The Kromix ads reportedly used benign cover images such as beaches, mountains, and waterfalls, masking the sexual content that followed. Meta’s head of women’s safety, Cindy Southworth, called this an adversarial tactic and said the company blocks many violating ads despite attempts to obscure them. Fair enough: fraudsters will always try to fool filters. But this is precisely why a system cannot be judged only by how many items it removes after the fact. Nudify app ads using deceptive thumbnails should face more than routine automated screening.
“Sexual exploitation, including promoting nudify apps, is horrific, and we work aggressively to fight it,” Southworth said.
Meta says it removed more than 340,000 violating nudification ads between late 2024 and early 2025, has introduced new detection technology, and has pursued legal action against some developers. That figure shows effort, but it also illustrates the scale of the market Meta is dealing with. If you pull 340,000 weeds from a garden, you don’t get to declare victory while the seed supplier is still open for business.
My read is that the incentives remain badly mismatched. Ad review has to move at commercial speed, and bad actors can cheaply create new accounts, revise a video, or swap a thumbnail. A newly created page with no followers running ads for an unknown AI app should trigger intense scrutiny, particularly when the product claims it can use real people’s faces. Instead, it seems the system still treats such campaigns as ordinary ad inventory until a researcher or journalist catches them.
Why female politicians are a particularly vulnerable target
Nonconsensual sexual deepfakes have targeted women for years, but AI image and video tools have made production faster, cheaper, and far less technical. Official portraits, campaign footage, and clips from legislative sessions are abundant online. They were once the routine visual record of democratic life. Now they can be ingredients for harassment.
Benjamin Shultz, lead researcher at the American Sunlight Project, described the shift plainly: “Formerly benign media—official portraits, video from a press conference—have become raw material for low- and no-code tools that anyone with an internet connection can use.” That should worry anyone who cares about elections, regardless of party. The goal need not be persuading voters that a fake video is real. Humiliation alone can work.
The threat is international, too. Public officials in Sweden, England, and Northern Ireland were among women targeted with sexualized AI imagery earlier this year. The issue has also become tangled up with general-purpose AI systems: reports around xAI’s Grok illustrated how quickly image-generation features can be turned into an industrial-scale harassment tool when meaningful constraints are absent.
That context makes nudify app ads involving political lookalikes especially grim. Platforms aren’t merely hosting user-created garbage after the fact; their ad systems can actively distribute invitations to create it. The difference is important. One is a moderation challenge. The other looks uncomfortably like paid customer acquisition for abuse.
Apple cannot shrug this off either
Kromix was reportedly available through Apple’s App Store before the inquiry that prompted scrutiny. Its paid features allegedly invited users to upload photos for scenarios with names including “bedroom rape.” That should be an obvious App Review rejection, not a close policy call.
Apple’s published App Review Guidelines prohibit overtly sexual material and apps that facilitate nonconsensual intimate imagery. Yet enforcement has been patchy enough that San Francisco’s city attorney sent Apple a cease-and-desist letter this summer over 13 named apps, accusing the company of helping enable their sale. Remember when Apple’s App Store pitch was that its tight controls made it safer than the open web? Cases like this put that claim under a very harsh light.
These two companies are connected by an awkward reality: harmful AI tools do not need a single giant platform to succeed. An app store can provide installation, a social network can provide customer acquisition, and payment systems can handle subscriptions. In this case, nudify app ads gave such tools a direct route to potential customers. Each company can point to the other and say it removed something when notified. Meanwhile, new variants arrive.
Removal is not the same as prevention
Meta says it shares information about removed apps and developers with other technology companies. That is sensible, and it may stop some repeat offenders. Still, nudify app ads are a test of whether platforms can identify the business model rather than merely chase individual creatives. A new developer, a scenic thumbnail, a vague AI label, and a link to an app should not be enough to reset the clock.
There is a practical answer here, even if it costs money: higher-friction reviews for ads selling face-generation or image-editing tools; verified developer identities; better detection of sexual content beyond a first-frame thumbnail; and rapid cross-platform bans for operators caught marketing nonconsensual imagery. None of that will erase the problem. But pretending this is an unavoidable glitch in automated moderation is no longer credible.
AI firms love to talk about democratizing creativity. Fine. But when the most profitable-looking use case is stripping women’s clothes digitally and publicizing the tool through paid ads, the industry has a much more basic question to answer: who, exactly, is doing the democratizing—and who is being made to pay for it?
