The apps that "show the face of your future baby" are image generators. They take two photos, blend features, and return a computer-generated portrait of a child. There's no prediction involved: the result is a montage, and another person with similar photos would receive a similar baby. As a couple's or group game, it works well. As a prediction, it doesn't exist.
Knowing this, what remains is what really deserves attention in this type of application: you are sending face photos — yours and someone else's — to a company's server you may never have heard of. This article deals precisely with that: what happens to these images, what Brazilian law says, and how to enjoy the fun without leaving biometric data scattered around.
How these apps work on the inside
There are two approaches. The older one uses image fusion: the program detects points on the face—eyes, nose, mouth, contour—calculates an average between the two faces, and applies the result to a baby model. It's the same technique as filters that swap faces in videos.
The latest version uses generative image models. From the photos submitted, the system produces a new face with compatible features. The result is more convincing, and precisely for that reason more deceptive: the child on the screen looks real, but was drawn pixel by pixel by an algorithm, with no relation to genetic inheritance.
A simple test reveals the nature of the thing: send the same pair of photos twice and compare the results. They usually come out different. If it were a prediction, it would be the same baby.
A photo of a face is sensitive data.
The General Data Protection Law classifies biometric data as sensitive category, ...with stricter treatment rules. Facial images used for identification fall into this category. In practice, this means that the service must clearly state what it collects, for what purpose, for how long it is stored, and with whom it is shared—and you have the right to request its deletion.
There's a point that almost no one thinks about: you're sending a photo of another person. The partner, the friend whose photo was included in the joke. That person did not agree to anything, and consent regarding the use of their image is theirs alone. It's best to ask beforehand.
What to look for before installing
- Where the processing takes place. Some apps process the image on the device itself; most send it to a server. The privacy policy should specify which is the case.
- Retention period. Look for the phrase that says how long the photo is stored after the result is generated.
- Used for training. Many terms include a broad license to use the submitted images to enhance the systems. If you don't want this, read it beforehand.
- Exclusion channel. There should be a clear way to delete the account and images, either within the app or via email to the data protection officer.
- Developer identification. Name, company, website, and other published applications. An anonymous developer with only one app is a bad sign.
- Recent reviews. This is where complaints about billing and unfulfilled deliverables first appear.
The subscription trap
The most common model in this category is a free trial of a few days that turns into a weekly or monthly charge. The typical sequence: you send the photos, wait for processing, the result appears blurry, and the print button leads to the payment screen.
If you decide to subscribe, note the date on your calendar and remember the rule: cancellation is done in the store, under "Payments & subscriptions" on Google Play or "Subscriptions" in your iPhone's Settings. Deleting the app does not stop any charges.
How to play safely
- Ask permission from the person who appears in the second photo.
- Choose a photo you've already posted, rather than a private image.
- Avoid using photos of real children in this type of service.
- Do not use social media login; if you do, revoke access afterward.
- Grant access to specific photos, not the entire gallery — Android and iOS allow this choice.
- When finished, delete the account through the app itself and uninstall it.
Extra care should be taken when publishing the results.
The generated image is of a child that doesn't exist, which already avoids most problems. Even so, posting the result along with the original photos exposes two real faces in the context of a "family in formation"—information that fuels fake profiles and social engineering scams. If you're going to share it, share only the result.
And a comment about pregnancy is in order: couples undergoing fertility treatment or in the adoption process sometimes become attached to these images. Since the portrait has no basis in reality, it can generate expectations that are not met. It's a joke, and it's better that way.
Where your photo goes after you tap send.
The path is almost always the same, and it's worth knowing because it defines the risk. Your image leaves your cell phone, travels across the internet to a server, is processed by machines with video cards—usually rented from a cloud provider—and the result returns via a file address. Along this journey, the photo passes through at least three hands: the application, the infrastructure provider, and, frequently, a third-party service that generates the image.
Two details often get overlooked. First: the result is hosted on its own address, and in many services this address is accessible to anyone with the link, even without an account. Sharing the link is different from sharing the image—the link remains active afterward. Second: Most of these servers are located outside of Brazil.. The General Data Protection Law allows for international data transfers, but requires that they be disclosed and that data protection guarantees exist. A privacy policy that doesn't specify where the data is processed has already failed the first test.
Add to that the backups. Even when the service deletes your photo from the main system, it can remain in backups for weeks until the rotation cycle expires. This is normal in any serious service — what's not normal is the policy saying nothing about it.
The photo carries more than just the face: the metadata.
Every photo taken with a cell phone carries an invisible record along with the file: exact date and time, device model, camera settings, and, if location tagging is enabled, the coordinates of the place where the photo was taken. A selfie taken in your living room provides your home address on this form with an accuracy of just a few meters.
When you send an image directly from your gallery to an online service, this metadata usually goes along. Some messaging apps remove metadata when compressing the photo, but sending it directly to a website or app usually preserves everything.
- Turn off location tagging on the camera. If you don't use this feature to organize your album, it's an option provided by the camera app itself.
- Remove the location when sharing. Android and iPhone offer this option on the photo sharing screen.
- Use a screenshot of the photo. When you want a quick way: the capture is created without the metadata of the original image.
- Look at the background of the image. A storefront number, license plate, ID badge, envelope with address, and school uniform say more than any coordinates.
You can change your password; but not your face.
A system that works with faces doesn't just store the photograph. It extracts a set of measurements—distances and proportions between points on the face—and transforms this into a numerical vector that allows for the comparison of two images. This vector is what makes the data so sensitive: it identifies you in other photos, on other services, without needing your name.
This is the key difference that sums up the issue. If a password is leaked, you can change it in five minutes. If a set of images of your face is leaked, there's no way to change it: your face remains the same for life. That's why Brazilian law places biometric data in the category requiring the strictest treatment, and why a prank service deserves the same care you would give to a document.
There is also a less discussed use: faces sent to image generators feed, in various services, the improvement of the models themselves — this is what the broad licenses in the terms allow. In practice, this means the possibility of producing new images with your likeness, in contexts you did not choose. This is not a reason to panic; it is a reason to read the clause that addresses the use of the images sent and not to pile up dozens of your photos on services you don't know.
How to request the deletion of your photos, in practice.
Requesting deletion is a right, and the request works best when written in the correct order. The script:
- Delete it first within the app. Look for the option to delete the account, not just the generated result. Deleting the image from the app's gallery rarely deletes what's on the server.
- Locate the contact channel. The store's privacy policy and data security sheet should include an email address for the data protection officer. The order will be sent to them.
- Write the request citing the law. The General Data Protection Law guarantees confirmation of data processing, access to data, and deletion. Request all three and provide the email or identifier used during registration.
- Ask for explicit confirmation. Find out if the images were deleted from the main system and backups, and ask if they were used for training or shared with third parties.
- Keep the answer. The law stipulates a period of up to fifteen days for a complete response to this type of request.
- If there is no response, File a complaint with consumer protection agencies and the national data protection authority. A service that ignores a deletion request is violating a legal obligation, and the complaint will be filed.
Photo of a real child: the line that doesn't cross.
There's a natural temptation with this type of app: to use a photo of a nephew, an older child, or a family child to "see how they would look." It's worth understanding why this is different from using your own photo.
Data concerning children and adolescents has its own specific regime under Brazilian law: processing must be done in the best interests of the child and depends on the specific consent of one of the parents or guardian. The Statute of Children and Adolescents expressly protects the image of minors. And, in practical terms, the child cannot consent and will not be able to request the deletion fifteen years from now, when that image may eventually be in some training set.
The recommendation is simple and has no useful exceptions: Do not send a photo of a real child to the image generation service., Whether it's your own or someone else's. If the curiosity is about family resemblance, the safest way is to compare old photos of relatives at the same age — it gives the same satisfaction and doesn't reveal anyone's face to any server.
Frequently Asked Questions
Will deleting the app delete the photos I've already sent?
No. Uninstalling removes the program from your phone and doesn't affect the data on the server. To delete the images, you need to delete your account within the service and then send a deletion request through the channel indicated in the privacy policy.
Is there a version that processes everything on the phone itself?
Yes, it exists, and it's the safest option in terms of data. A practical way to check: put your device in airplane mode and try to generate the result. If it works without internet, the processing is local. If it gives an error, your photo was being sent abroad.
Can the service use my image in an advertisement?
It depends on what you agreed to. Many terms include a broad license to use the submitted material for the dissemination and improvement of the system. This clause is usually written out, and it's the one worth looking for before submitting anything.
How does an app like this make money if the trial is free?
Through three paths, usually combined: recurring billing after a short period, advertising displayed between screens, and the value of usage data itself for ad targeting. Image generation is expensive processing—when there's no visible price, the revenue is elsewhere.
Is logging in with a social media account safer?
It's more convenient and less private: besides eliminating the need for a new password, it gives your name, photo, and sometimes your contact list to the service. Prefer registering via email. If you've already used social login, revoke access in the list of connected applications on your network.
Conclusion
This app, which supposedly shows the face of a future baby, is purely for visual entertainment: it generates a plausible image from two photos and doesn't predict any real appearance. Just show the same pair of photos twice, and the app itself will provide proof.
What truly deserves attention is what you give in return: facial images, which the law treats as sensitive data, often of a person who wasn't even consulted. Ask for permission, read the privacy policy, grant access only to the chosen photos, take care of the signature, and delete the data when finished. The fun continues just as much—and without leaving a trace.
Read too
- View Your House by Satellite: What Appears and How to Request the Blur
- What Will Your Baby Look Like? What Genetics Says and What the App Does
- LGBTQI+ Chat: Chat Apps and Privacy Policy
- GPS that avoids tolls and traffic: When it's worth it and when it's expensive.
- Shared Wi-Fi Password: How the Collaborative Base Works
- Asian Films Dubbed: How to Find Where Each Title Is Available
