How to tell if an image is AI: questions and answers

The most reliable way today is to run the image through an AI image detector and check where it came from. Visual clues still help: warped text, odd hands, earrings or glasses that don't match, shadows that disagree with the light.

But the newest generators rarely make those mistakes. WAAIV checks the picture with several specialized AI detectors at once and tells you, in plain language, how likely it is to be AI.

Because modern AI image generators no longer leave obvious mistakes. Diffusion models learn from billions of real photos and build each image pixel by pixel, so skin, light and background blur look natural.

On top of that, almost every picture you see online has been resized and recompressed by Instagram, WhatsApp or X, which erases fine details and usually removes the metadata (EXIF, Content Credentials) that could reveal where it came from. And each generator leaves different traces: detectors trained on older models tend to do worse on new ones. So the human eye, and even a single detector, can be fooled.

Most AI image detectors are machine-learning models trained on millions of real and AI-generated images. They learn patterns people can't see: how noise is spread across the image, fine textures, how edges and colors are formed.

Some also read the file's metadata or look for invisible watermarks such as Google's SynthID or C2PA Content Credentials. Those help only while the file still carries them, and social media often strips them. The output is a probability, not a simple yes or no.

Yes. No AI detector is right 100% of the time. Real photos can be flagged when they are heavily edited, filtered, very small or compressed many times, and brand-new generators can slip past a detector that hasn't learned them yet.

In a May 2026 test by NewsGuard, five popular detectors marked real news photos as AI about 13% of the time. That is why WAAIV never relies on a single detector, and why the result should be read as strong evidence, not proof.

Be careful before sharing it. Look for the original source: who posted it first, when, and whether trusted media show the same scene.

If it is about the news, check a fact-checking site. And for anything important, don't rely on one signal alone, ours included.

Because WAAIV combines several leading AI detection tools instead of trusting just one. Each image is checked by several detectors at the same time, and each one is strong in different situations.

We test them against images whose origin we know and give more weight to the ones that get it right more often. As new AI generators appear, we keep reviewing the tools and update the mix when a better one comes along. Then an AI explains the result in plain language, so you understand why, not just a number.

Yes. That is one of the things that makes WAAIV different from other AI image detectors: after each scan, an AI explains the result in plain language, and you can ask it anything about that analysis. For example: why the detectors disagree, how to explain the result to someone who sent you the image, how to find where the photo came from, or what to check before sharing it.

It works on the website with a free account, including for past scans in your history, and on WhatsApp (just write in the chat after the result). Asking is free: questions don't use up your scans and cost you nothing. The AI answers from the detectors' results: it helps you understand them, but it doesn't replace checking where the image came from.

Yes, with a free account you have a history of everything you checked. Each scan shows the image, the result and the explanation. Scans you make on WhatsApp appear there too once your number is linked to your account.

You can open any past scan to ask more questions about it, or remove it from your history.

WAAIV tells you whether the image is likely real or AI-generated, whatever tool was used. Pictures from Midjourney, ChatGPT, Gemini, Stable Diffusion, Flux and other generators all go through the same check.

Guessing the exact tool is much less reliable than detecting AI itself, so we don't present it as a fact.

It depends on how big the edit is. WAAIV is built to spot images created by AI, and large AI changes usually show up.

Small touch-ups, like a filter, a crop, smoother skin or a removed object, often don't, because most of the photo is still real. If you suspect one part of a photo was changed, also look for the original version.

Yes. You can check photos, screenshots and images saved from Instagram, Facebook, X, TikTok or WhatsApp, up to 15 MB.

Tip: use the best-quality version you can find. Every time an image is shared it loses detail, which makes any analysis harder.

Yes, to start. On the website your first scan is free, with no sign-up. A free account gives you more: +1 with email or +2 with Google, up to 3 free scans in total. On WhatsApp you get 3 free scans.

Free scans are a one-time allowance per person; they don't reset every day. If you need more, you can buy a pack of scans.

Yes. When your free scans run out, you can buy a pack of 10, 30 or 100 scans, with no subscription. Create a free account and choose a pack on the website. Your scans work on the website and also on WhatsApp once your number is linked to your account.

Our prices are low. Today you can pay in Colombian pesos by card, PSE, Nequi, Bancolombia or Daviplata, and we are adding more payment methods step by step.

Yes. Send or forward the image to WAAIV on WhatsApp and the answer arrives right in the chat. You can also ask follow-up questions about the result.

For now it works for phone numbers from Colombia and Brazil, and more countries are coming soon. Anywhere else, you can use the website.

We use your images only to give you the result. To analyze them, we send them to the AI detection providers we work with. We don't use them for advertising and we don't sell them.

You can ask us to delete your data at any time. Details are in our Privacy Policy.