NoPeek

Inspiration

In today’s world, photos travel online at lightning speed—often before we even think about what’s in them. A simple selfie could expose your location through hidden EXIF data, while a quick street shot might reveal a stranger’s face or a license plate without their consent. Social media brings us together, but it also blurs the boundary between sharing and oversharing.

This is where AI proves its worth—not as a trendy buzzword, but as a practical tool. With intelligent detection and creative cartoonization, AI can automatically identify sensitive details and give you a safer, more anonymous way to share.

So we asked: what if privacy wasn’t something you added after the fact? What if protecting yourself online was as effortless as hitting “post”?

That idea sparked NoPeek—a privacy-first solution built to let you share confidently, without revealing more than you intend.

What It Does

NoPeek is a smart privacy assistant that works quietly behind the scenes—before your content hits the timeline.

Say you snap a photo and get ready to upload it to Instagram, TikTok, or X. Before you even post, NoPeek jumps in to:

Scan the image for faces, license plates, documents, EXIF metadata, and more
Automatically apply privacy filters using AI

What kind of filters? You choose:

Turn real faces into cartoon avatars
Apply blur to sensitive areas
Cover things with fun stickers
Decide who gets to see what—maybe your close friends see the original, but followers see a protected version

NoPeek doesn’t just flag privacy risks—it solves them.
And best of all, it adapts to your rules, every time you post.

How We Built It

We built NoPeek from scratch with real-world usage in mind—speed, reliability, and plug-and-play integration were key.

Here’s what’s under the hood:

Frontend:

We use the Lynx Framework to build cross-platform applications, which greatly reduces our team’s workload while leveraging the existing web development expertise of our members. With the help of Lynx Native Modules and Native Elements, we have successfully built Bridge to access native functionalities such as ImagePicker, PhotoLibrary, FileManager, and GestureRecognizer. This enables our Lynx applications to deliver a smooth and familiar user experience.

In terms of interaction design:

We designed a Comparator that allows users to compare the original and the processed images.
We implemented a Slider to visualize the processing workflow, helping users keep track of the current stage.
During the EXIF-removing process, a short animation is played so that users experience the operation seamlessly without feeling the wait.
At every step of the process, we provide options like “Skip Process” or “Select Which to Go”, giving users full control over how their photos are handled.
We built a Gallery that records the images users have processed. From here, they can share masked images to publishing platforms, save them to the Photos Album, or revisit and check which parts of the image were processed.

Metadata Scrubbing

We use tools like piexifjs to wipe hidden data like GPS, camera info, and timestamps

Visual Detection

We combined YOLO for license plates, InsightFace for facial detection, and custom computer vision pipelines for documents and other private content

Avatar Generator

For cartoon-style anonymization, we tap into the FLUX diffusion model, which turns real faces into stylish, creative alternatives

Every tech choice was made with real users in mind—fast enough to be seamless, accurate enough to be trusted.

Challenges We Ran Into

NoPeek didn’t build itself—and the road wasn’t exactly smooth.

Learning Lynx on the fly: The framework we chose was powerful, yet unfamiliar to our team. We had to figure it out fast—and build on top of it even faster.
Backend from scratch: From managing uploads and queues to storing user settings securely, we had to architect a scalable system on tight deadlines.
Research vs. reality: Many AI models look great on paper. But we had to balance speed, memory footprint, and reliability to find the few that worked well in the wild.

In short, we had to turn academic ideas into something real people could actually use.

What We’re Proud Of

Despite the time crunch, we delivered something real—and really useful.

Some of our proudest wins:

Built a complete privacy pipeline from detection to obfuscation to sharing—end-to-end
Got Lynx integration working with platforms like TikTok so privacy protection feels like part of the app, not an afterthought
Made privacy feel personal and transparent—users can see exactly what’s being hidden and tweak it to their needs

It’s not just about saying you value privacy. It’s about showing up for it, every time someone posts.

What’s Next for NoPeek

We’re just getting started. Here’s where we’re headed:

Currently, we have successfully built an iOS app (xcarchive) that is ready to be deployed to the App Store.
We plan to provide broader cross-platform support by leveraging the powerful capabilities of Lynx to build applications for iOS, Android, and the Web.
Smarter detection of IDs, passports, credit cards—anything sensitive in a frame
Introducing @Privacy: tag someone, and NoPeek will apply their privacy preferences automatically
Giving users finer control: show your dog clearly, blur your coworker; hide your location, but keep the vibe
More styles beyond blur: thermal overlays, silhouettes, adaptive masking, and more
For desktop users, we are considering developing a Browser Extension to provide more seamless experience.