fake news & propaganda
the eiffel tower is on fire?
it burned down on your feed this morning, 40k reposts deep. amige reads the pixels likely ai, and no camera-signed credential shows up to argue otherwise. paris is fine.
amige: AI image & video detector. check if an image, video or text is AI-generated. is it real? or is it?
image, video, or text, amige. reads it against
every model that matters, names what made it,
and never fakes certainty.
drop a photo, video,
or chunk of text.
or chunk of text.
or paste text instead↓ instant verdict
// tungtung.jpg
likely ai99%
// starry_night.jpg
likely human92%
// T-800.jpg
uncertain/
AI enhanced?54%
AI enhanced?54%
// 💀.mp4
likely ai99%
← swipe to see more →
what amige. does
90+generative models,
one place
one place
one scan, any medium
image, video, or text — checked against 90+ generative models in one place.
looks like midjourney v6
best guess · 71%
gpt-4o image22%
gemini · imagen7%
names the maker
not just “it’s AI.” amige. says which model it resembles, when it’s confident.
a panel, not a guess
amige. routes each scan to the detectors strongest for it, then fuses them into one verdict. disagreement shown, never averaged away.
71%
never 0%uncertainnever 100%
honest by design
every read is a probability, never a proof. it tells you how sure it is, and says “uncertain” rather than guess.
many minds.
one honest call.
a scan is read, routed, weighed by a panel, then fused & calibrated — landing on a probability, never a proof. watch one go through.
how it works, in full →↳ one scan, start to finish.
↳ one scan, start to finish.
↳ one scan, start to finish.
How an amige. scan works, step by step. First, the content you drop — an image, video, or text — is read for signal. Before any verdict, amige. takes two quick looks that guide the scan but never score it: a routing hint that points toward the detectors most likely to read this kind of content well, and a novelty check that notices when something is unusual so the engine treads more carefully. The novelty check only flags that the content is unusual; it never claims what made it, and it is not a new-generator detector. Next, a panel of independent detectors — built by different teams on different data — each weigh in on their own. They do not all agree, and the disagreement is kept on display, not averaged away. Their readings are then fused and calibrated into a single estimate: combining only tempers confidence, it never inflates it, and the result is capped so a verdict never reads 0 percent or 100 percent. The outcome is one of three honest calls — likely AI, likely human, or uncertain — always a probability, never a proof. When the panel splits or the content is unusual, amige. holds the verdict back as uncertain rather than guess. The readings shown in the animation are illustrative of a single scan and are not accuracy claims.
it's real?
or is it?
← image or amige? there's only one way to find out.- 01drop itimage · video · text. up to 15mb. heic + mov welcome.
- 02let amige. detectlikely ai or likely human, across 90+ models from midjourney to gpt-4o to nano banana.
- 03every model, every numbernot an average. ranked side-by-side so you see the reasoning, not just the verdict.
catfishing & romance scams
the match at the trevi fountain.
matched monday, rome selfie by tuesday. perfect skin, perfect light, no such person. amige reads the face likely ai before you start learning italian.
deepfakes & impersonation
the video call that feels a half-beat off.
the mouth lands late and the blink comes in wrong. amige runs the clip frame by frame and reads it likely ai. hang up, call back on the number you already had.
housing & rental scams
the listing that’s somehow half price.
a dumbo loft with the brooklyn bridge filling the window, at half the going rent. no landlord took this photo. amige reads the listing likely ai. keep your deposit.
travel scams
the santorini suite at motel money.
cave walls, plunge pool, the caldera out the open door, and a nightly rate that makes no sense. amige reads the listing likely ai before the “host” stops answering.
e-commerce scams
the drop that never ships.
a grail colorway back at retail price at 2am. zoom on the wings logo: the letters spell nothing. amige reads the shot likely ai before checkout does the damage.
art theft & style mimicry
a “lost van gogh,” minted last night.
same swirls, same stars, no van gogh. a model can wear an artist’s whole style. amige reads the brushwork likely ai and names the tool it resembles. receipts for the group-chat art historians.
fake reviews & ai text
the reviews that closed the sale.
five stars, zero humans. pangram, winston and ai-or-not all land on likely ai. and amige says when it isn’t sure: non-native english reads “too clean” too.
it's real?
or is it?
← image or amige? there's only one way to find out.- 01drop itimage · video · text. up to 15mb. heic + mov welcome.
- 02let amige. detectlikely ai or likely human, across 90+ models from midjourney to gpt-4o to nano banana.
- 03every model, every numbernot an average. ranked side-by-side so you see the reasoning, not just the verdict.
fake news & propaganda
the eiffel tower is on fire?
it burned down on your feed this morning, 40k reposts deep. amige reads the pixels likely ai, and no camera-signed credential shows up to argue otherwise. paris is fine.
catfishing & romance scams
the match at the trevi fountain.
matched monday, rome selfie by tuesday. perfect skin, perfect light, no such person. amige reads the face likely ai before you start learning italian.
deepfakes & impersonation
the video call that feels a half-beat off.
the mouth lands late and the blink comes in wrong. amige runs the clip frame by frame and reads it likely ai. hang up, call back on the number you already had.
housing & rental scams
the listing that’s somehow half price.
a dumbo loft with the brooklyn bridge filling the window, at half the going rent. no landlord took this photo. amige reads the listing likely ai. keep your deposit.
travel scams
the santorini suite at motel money.
cave walls, plunge pool, the caldera out the open door, and a nightly rate that makes no sense. amige reads the listing likely ai before the “host” stops answering.
e-commerce scams
the drop that never ships.
a grail colorway back at retail price at 2am. zoom on the wings logo: the letters spell nothing. amige reads the shot likely ai before checkout does the damage.
art theft & style mimicry
a “lost van gogh,” minted last night.
same swirls, same stars, no van gogh. a model can wear an artist’s whole style. amige reads the brushwork likely ai and names the tool it resembles. receipts for the group-chat art historians.
fake reviews & ai text
the reviews that closed the sale.
five stars, zero humans. pangram, winston and ai-or-not all land on likely ai. and amige says when it isn’t sure: non-native english reads “too clean” too.
the gallery.
real scans from our users, click any for the full breakdown ↓
← swipe to see more →
// tomcruise_temu.mp4
deepfake/
AI face-swap detected99%
AI face-swap detected99%
2 detectorsdreamid
// miule.jpg
likely human86%
2 detectorsno model match
// monalisa.jpg
likely human86%
2 detectorsno model match
// gptdoodles.jpg
uncertain/
AI enhanced?45%
AI enhanced?45%
2 detectorsgpt-4o
// midterm_thesis.pdf
likely ai100%
2 detectorsai-written text
// ronaldo_messi_3.mp4
deepfake/
AI face-swap detected84%
AI face-swap detected84%
2 detectorsgrokimagine
we wrote it down ↓browse all →
methodology
a panel of detectors.
a panel of independent detectors, fused into one honest verdict. how amige.’s ensemble fails less catastrophically than any single classifier. limitations included.
read →
glossary
what’s a deepfake?
synthetic audio, image, or video that depicts a real person. three subtypes, 2024 election context, the harm that’s actually rising.
read →
model guide
how do you spot ChatGPT?
the highest-traffic AI tool in history. GPT-3.5 through GPT-5, the writing tells that give it away, and how amige. catches it.
read →
comparison
which detector wins?
no single tool wins every case. the image, video, and text detectors ranked by what they’re actually good at — bias disclosed, weaknesses named.
read →
how-to
test it before you trust it.
“99% accurate” is doing a lot of work. a 30-sample blind protocol — human, AI, and hybrid — to check any detector’s claim yourself, ours included.
read →
video guide
is that a Sora video?
physics first, watermark second. the morphing backgrounds, off-beat hair, and liquid that behaves wrong — the tells that give OpenAI’s video model away.
read →
questions? good.
what is amige.?
amige. is an AI content detector. drop in an image, video, or text and it detects whether content is likely AI-generated or human-made, then names the most likely model behind it, like ‘looks like Midjourney v6’.
is amige. free?
yes. there is a free tier of 3 scans a day, and you can run one without an account. paid plans add higher image and video limits plus unlimited text.
how accurate is amige.?
amige. runs a panel of independent detectors and shows where they agree or disagree, with a per-detector breakdown. no detector is perfect, and accuracy drops on heavily edited or re-shared media, so amige. surfaces the reasoning behind each verdict.
can amige. detect deepfakes?
yes. amige. runs deepfake-specific checks on images and video alongside the general AI-generation detectors, and flags the most likely manipulation type in the result.
doubt is the point ↗all 16 questions →