Deepfake App Removal: Why the Fight to Contain Them Is Just Starting

The app was rated "E for Everyone." You could find it by typing a few words into the App Store search bar. It wasn't hidden in some dark corner of the internet. And it could take a photo of a real person, your neighbor, your coworker, your teenager's classmate, and generate a fake, sexualized image of them in seconds. San Francisco's City Attorney has now sent cease-and-desist letters to both Apple and Google, demanding they pull these apps. The apps had been flagged by researchers for nearly a year before legal threats finally made the companies move.

This isn't a story about celebrities getting deepfaked. It's about tools designed to harm ordinary people that were available on the most trusted digital storefronts on earth. If you've ever posted a photo of yourself online, or let your kid post one, this story is about you.

---

Nudify Apps Were Visible in Plain Sight

Here's what makes this story different from the usual deepfake headlines. We're not talking about shadowy websites or illegal file-sharing servers. The apps in question, names like PicX, Tapart, MatureAI, and Artifusion, were inside the official App Store and Google Play. Some disguised themselves as ordinary photo-editing tools. One was rated appropriate for all ages. According to Engadget, at least some of them showed up in basic keyword searches, no password, no secret handshake required.

The Tech Transparency Project, a nonprofit research group that investigates big tech, published findings in both January and April of 2026 identifying dozens of these apps. They sent their findings directly to Apple and Google. Both companies kept the apps up. Both companies kept collecting their cut of every transaction those apps processed. It wasn't until San Francisco's City Attorney sent formal cease-and-desist letters that meaningful action started.

That timeline matters. Nearly 365 days of documented awareness. Zero meaningful response. One legal threat. That's not a detection problem, that's a choice.

"Apple and Google have likely made millions of dollars in fees from apps offering these services." San Francisco City Attorney's Office, as reported by TechCrunch

Read that again. The platforms didn't just fail to notice. They profited. Every time someone paid for a "premium" nudify feature, Apple or Google took a standard cut, usually around 15 to 30 percent. That's not a bug in the system. That's the system working exactly as designed, just pointed in a terrible direction. This article is part of a series, start with Your Face Was Scanned Saturday Nobody Asked If That Was Lega.

---

Why This Is Everyone's Deepfake Safety Problem

There's a version of this story that feels distant. Famous person gets deepfaked. Headlines for a week. Congressional hearing. Nothing changes. We've watched that loop play out before.

This is not that story.

More than one in three teenagers. Not celebrities. Not public figures. Kids. And 33.2% reported having a non-consensual image like this shared without their knowledge. These numbers come from a peer-reviewed study, the kind that gets reviewed by other scientists before it's published, not just an advocacy group running a poll.

What nudify apps did was collapse the barrier between "this could happen to a famous person" and "this could happen to anyone with a social media profile." You don't need a high-resolution photo session. A selfie will do. A school picture. A LinkedIn headshot. The AI fills in the rest, and what it fills in is a lie that looks real enough to destroy someone's reputation, relationship, or sense of safety.

If you're a parent and you just felt your stomach drop, that's the right reaction. Researchers at RAINNone of the country's leading sexual-violence organizations, have documented the legal gaps that let this harm fall through the cracks. The harm is real. The tools are real. And until this week, the tools were available in the same app store where your kids download games.

---

Apple and Google Missed Clear Warning Signs

The obvious pushback from Apple and Google goes something like this: we review millions of apps, bad actors disguise themselves, we can't catch everything. And honestly? For a lot of abuse cases, that argument has some weight. Previously in this series: Your Face Got Scanned 2 600 Times A Court Just Said Thats On.

Not this one.

The apps documented by the Tech Transparency Project, and catalogued in detail by NCOSE (the National Center on Sexual Exploitation), weren't subtle. Several advertised their actual purpose in their descriptions once you dug past the "photo editor" branding. Researchers handed Apple and Google specific app names, payment evidence, and reports showing these companies were actively steering users toward the tools. At that point, "we didn't know" becomes very hard to say with a straight face.

What You Can Actually Do Right Now

Okay. You've read the scary part. Here's the part where you get something useful.

First: do a quick audit of what photos of you, and your kids, are publicly visible. Not "does my Instagram account exist" but "can a stranger download a clear, well-lit photo of my face from it?" If the answer is yes, consider tightening those privacy settings. This doesn't mean disappearing from the internet. It means making the casual grab harder.

Second: if you have teenagers, have the direct conversation. Not the vague "be careful online" talk, the specific one. Tell them that AI can take a photo of anyone and generate fake images that look real. Tell them that if it ever happens to them or a friend, it is not their fault and it is not legal. Tell them to come to you. Kids are far more likely to handle something like this well when they know the adult in their life already knows it exists.

Third, and this is the one that surprises people, start paying attention to whether a photo or profile claiming to be a real person actually checks out. If you've ever looked at a dating profile or an online contact and thought "is this person actually who they say they are?", that instinct is exactly right to trust. Tools that verify real identity from photos exist precisely because the fake-image problem has gotten this serious. Knowing the difference between a real face and a generated one isn't paranoia anymore. It's just being informed. Up next: Monroe County Biometric Disclosure Retail Facial Recognition.

---

The Labeling Debate Is a Distraction

Some people will argue that the answer here isn't removal, it's warning labels. Slap a disclaimer on the app. Make users confirm they're over 18. Problem solved, right?

No. And here's why that argument falls apart immediately: a warning label on a weapon is not the same as locking up the weapon. These apps don't require a victim's consent to use. They don't require the person in the photo to know anything is happening. A label that says "don't use this to harm people" means exactly nothing to someone who opened the app with the intention of harming someone. The only thing that stops a tool built for abuse is removing access to it.

The San Francisco City Attorney gave both companies a 28-day deadline to comply. That's a short window. Whether Apple and Google treat it as a real deadline or a public-relations timeline to manage will tell us a lot about whether platform safety is something they actually care about, or just something they say they care about when a lawyer is watching.

Here's the thing that will stay with me about this story: these apps passed the review process that's supposed to protect us. They wore a disguise, "photo editor," "E for Everyone", and walked right in. Which means the safeguard we've been trusting to keep our phones safe was, in this case, the thing that made the harm easier to reach. That's not a small glitch. That's a design question that neither company has answered yet.

Your photo is not permission. It never was. But right now, somewhere in a developer's app submission queue, there's probably another version of these tools waiting to try again.

What Exactly Is a Deepfake App?

A deepfake app is any piece of software that uses artificial intelligence to alter a real photo or video so convincingly that it looks genuine, even though it never happened. Most deepfake apps work by feeding a real image into a model trained on thousands of faces, then using that model to swap, stretch, or fabricate details the original photo never had. The apps named in San Francisco's cease-and-desist letters fall squarely into this category, they are deepfake apps built specifically for face swapping and generating sexualized images without consent.

Not every deepfake app is harmful by design. Some are marketed as entertainment, letting people put their own face into a movie scene or swap faces with a friend for a laugh. But the deepfake apps at the center of this story were not built for jokes. They were built to strip a real photo and generate a fake nude, and that single purpose is what separates a harmless novelty from a tool of abuse.

How a Deepfake Generator Actually Works

Every deepfake generator relies on the same basic idea: show the AI enough real faces and enough real bodies, and it learns to generate a convincing fake body attached to a real face. The mobile app doesn't need a studio or a technical team behind it, a single photo uploaded from a phone is often enough input for the deepfake generator to produce a finished image within seconds.

This is part of why the harm scales so quickly. A face swap that once required real skill and expensive software now happens inside a mobile app that costs a few dollars, or nothing at all. Online deepfake tools have gotten so efficient that the technical barrier that used to protect people has essentially disappeared.

Deepfake Detection Is Still Playing Catch-Up

Deepfake detection technology exists, and researchers are actively improving it, but it has never kept pace with how fast deepfake software spreads. A detection tool built to catch last year's face swap techniques often fails against this year's deepfake technology, because the underlying models keep getting better at producing realistic video and photo results. That gap is exactly why relying on detection alone, after the fact, isn't a real solution, removing the deepfake apps before the harm happens matters far more.

Platforms have talked about building detection directly into their review process, scanning new app submissions for signs of nudify or faceswap functionality before approval. That kind of proactive deepfake detection would catch a tool like this before it ever reaches a search result, instead of relying on outside researchers to do the work for free.

Face Swapping Isn't New, But the Speed Is

Face swapping as a concept has existed in film and photo editing for decades, special effects teams have swapped faces in movies using far more labor-intensive methods long before AI made it simple. What changed is speed and access. Face swap technology that once required a visual-effects budget now runs on a mobile app anyone can download in under a minute.

That shift from expensive and slow to instant and free is the real story behind every headline about a deepfake app. A face swap used to be a specialized skill. Now it is a checkout button.

Video is where this problem grows next. A single photo used to be the input for most of these tools, but deepfake software that can generate a moving, talking video from just a few still images is already spreading. If a fake photo can damage a person's reputation, a fake video that appears to show them speaking or moving is even harder for a viewer to dismiss as fake, and even harder for the person targeted to disprove.

Parents and schools should treat videos with the same skepticism people are slowly learning to apply to photos. A convincing video clip shared in a group chat carries an authority that a still image doesn't, simply because moving images have historically been harder to fake. That assumption no longer holds, and the sooner people understand that any video could be a deepfake generated in minutes, the less power these fake videos have to cause harm.

None of this means every photo-editing mobile app on your phone is dangerous. Most are exactly what they claim to be. But the lesson from San Francisco's action is that "available in an official app store" was never a guarantee of safety, and it never will be until platforms build real deepfake detection into their review process instead of waiting for outside pressure to act.

Where Deepfake Apps Actually Get the Face

Every deepfake app needs a starting photo, and most people never realize how little the app actually requires. A single face swap can be generated from a public profile picture, a school yearbook scan, or a tagged photo from a birthday party. The maker of the app doesn't need a full photoshoot, one clear, front-facing image is often the only raw material a face swap tool needs to create a fake image.

This is why "I never post revealing photos" isn't the protection people think it is. A completely ordinary photo, fully clothed, smiling, taken at a family event, is enough content for a deepfake app to swap onto a fabricated body. The face is the only ingredient these tools actually care about, which means the pool of possible targets is basically everyone with a public photo online.

Why "Free" Deepfake Tools Aren't Really Free

Many of the apps at the center of this story advertised themselves as free deepfake tools, at least at first. A free version usually creates a single low-resolution image, just enough to hook a user, before pushing them toward a paid tier for a sharper or more realistic video. That's the business model: free gets people in the door, and Apple and Google collect a cut once someone pays to unlock the fuller version.

There's also a cost that has nothing to do with money. Creating deepfake content, even through a free maker with no subscription attached, still requires uploading someone's face to a server the victim never agreed to and never sees. The word "free" describes the price to the person using the app. It says nothing about the cost to the person whose photo got turned into content without their knowledge.

What "Realistic" Actually Means Here

When people call a deepfake image or deepfake video realistic, they usually mean it's good enough to fool a quick glance, a parent scrolling a group chat, a classmate seeing a shared screenshot, a stranger reposting it without checking. That's a much lower bar than "indistinguishable under close inspection," and it's the bar that actually matters for real-world harm. A fake image doesn't need to survive forensic analysis to wreck someone's day, their reputation, or their sense of safety.

This is also why relying on people to just "tell" whether an image is fake has never been a workable safety plan. As AI tools have improved, the visual glitches that used to give away a fake, strange hands, mismatched lighting, blurred edges around the face, have mostly disappeared. Realistic used to mean rare. Now it's the default setting.

None of these details are meant to be a how-to guide. They're meant to explain why removal, not detection or disclaimers, is the only response that actually protects people, because once a deepfake app can create a realistic fake from a single ordinary photo, the damage happens the moment the image is generated, not the moment someone finally proves it's fake.