In a world where fakes are easier to make than ever, trust is no longer something we can afford to take at face value—especially when it comes to digital screenshots. From forged payment receipts to manipulated social media messages, screenshots are becoming one of the most abused tools for misinformation. That’s what led me to create Screenshot Sleuth GPT, a custom AI assistant designed specifically for fake screenshot detection.
Why I Built Screenshot Sleuth GPT
It started with a simple observation: fake GCash receipts were being passed around, and some of them looked scarily convincing. Although it’s easy enough to check if a payment went through by looking at your balance or transaction history, people still fall for these image-based scams—especially in buy-and-sell groups or informal online transactions.
Out of curiosity, I even tried generating a fake GCash screenshot using ChatGPT. While it wasn’t flawless, it was easy to imagine someone falling for it—especially at first glance. But a closer inspection revealed telltale signs of manipulation. AI-generated images tend to introduce small but noticeable artifacts: the woman in the ad looks off, and tiny text often breaks down into gibberish or glyphs. Worse, I realized that using Photoshop was even more effective. You could simply adjust a few numbers and export it with minimal effort.
My thought process was simple: if scammers are starting to use AI tools to generate fake receipts, then there should be an AI tool that helps people avoid falling for them. I figured, if there’s a tool that can provide a second opinion on a suspicious screenshot, that might just help someone avoid a scam.
So Screenshot Sleuth GPT was born. It’s not meant to replace common sense or direct account verification. Instead, it offers a layer of scrutiny focused on fake screenshot detection, looking for signs of tampering, image compression artifacts, inconsistencies in layout, and metadata anomalies.
What Screenshot Sleuth Can Do (Right Now)
At its current stage, Screenshot Sleuth GPT is primarily focused on analyzing uploaded images for signs of manipulation. It uses a combination of visual inspection cues and metadata analysis to flag irregularities. For example:
- Are there signs of cloning or digital editing in certain areas?
- Does the metadata contradict the visual content (e.g. creation date vs. timestamp on the screenshot)?
- Are there compression artifacts or image glitches that seem out of place?
While I have prompted it to not specifically analyze UI/UX elements like layout design or branding accuracy, it does highlight visual inconsistencies that might be indicators of tampering. It can’t guarantee 100% accuracy, but it can raise enough red flags to make you think twice. It’s a lightweight but helpful assistant for fake screenshot detection.
Where Screenshot Sleuth Is Headed
In the future, I want to expand Screenshot Sleuth GPT beyond just image analysis. The goal is to evolve it into a more general-purpose verification assistant—a digital detective that can:
- Cross-reference screenshots with live data or known formats
- Analyze viral screenshots of “news” or quote cards for factual accuracy
- Link up with third-party databases or services for quicker verification
I’ve already begun refining its prompts and natural responses, but there’s more in store. Planned backend upgrades include:
- An AI vs. Real detection system
- The ability to return an ELA (Error Level Analysis) overlay preview via URL
- Improvements to its analytical engine
On the user experience front, I’m working toward:
- Reverse image search links
- A side-by-side comparison tool for two screenshots
- Better user onboarding and support features
Further down the road, I aim to implement broader fake news detection capabilities—transforming Screenshot Sleuth into a full-fledged misinformation watchdog. But at its heart, its primary mission will remain fake screenshot detection.
Think of Screenshot Sleuth as Snopes reimagined: faster, GPT-powered, and image-aware.
A Work in Progress, But With Purpose
I won’t pretend this GPT is perfect. It’s still learning. And there are real limitations to what image analysis can tell you. But this is a first step toward building a tool that fights misinformation in the medium where it thrives: visual evidence. We can’t stop people from faking screenshots, but we can give others the tools to catch them.
If you’ve ever second-guessed a suspicious payment receipt or been burned by a fake confirmation message, Screenshot Sleuth GPT is here to help you double-check—one pixel at a time. It’s one more tool in your digital safety kit for fake screenshot detection.
To learn more about how Screenshot Sleuth respects your privacy, you can read the Privacy Policy for full details on how your inputs are handled and protected.
About Me
I’m JP B. Bantigue, CLSSBB—a UI/UX designer and digital creative who believes technology can and should serve truth. Screenshot Sleuth is my attempt at bringing a layer of digital forensics to everyday users, making the internet a little safer, one screenshot at a time.
