The quiz loads in this browser.
Free · In this browser · No AI call
Real Photo or AI
Look at one picture and say whether a camera or an AI made it. The real photos are public-domain or CC0 pictures of a forest, a cup of coffee, an alley, a fox, a coast, a loaf, a field of stars, a desk, a bee, and a barn. The AI pictures cover the same subjects and were generated for this page. A round is 10, shuffled, five of each. The reveal names the source, then one or two habits to notice. Detection is unreliable, and the page says so.
- Look One picture. The description does not give the answer away.
- Choose Real or AI. Press R or A, or use the buttons.
- See the source Author and license, or a line that says the picture was generated by AI for this site.
Photo 1 of 10
Best on this device: none yet.
Your eye accuracy
No answers yet
Press R for Real or A for AI. The buttons do the same thing.
Credits
The real photographs below are public domain or CC0. The source link is the Wikimedia Commons file page where that license is recorded. Thumbnails are the same files the quiz uses, stored on this site. Nothing is fetched from Commons, Flickr, NASA, or the National Archives while you play. The ten AI pictures (forest, coffee, alley, fox, coast, bread, stars, desk, bee, and barn) were generated for this site. They are not from these archives, and the reveal labels them Generated by AI for this site.
-
Nurse log on the Hoh Trail, Olympic National Park. Olympic National Park, National Park Service. Public domain, U.S. government work (PD-USGov-NPS). https://commons.wikimedia.org/wiki/File:Nurse_Log_Hoh_Trail_rain_forest_NPS_Photo_MP1_(23074100775).jpg
-
Cup of coffee on a wooden table. Wendelin Jacober. CC0 1.0 Universal. https://commons.wikimedia.org/wiki/File:Another_Cup_Of_Coffee_(_Public_Domain_).jpg
-
Chinatown alley, San Francisco, circa 1906. Unknown photographer, circa 1906. Public domain (PD-US). The Bancroft Library holds the scan, digital ID cubcic brk7484. https://commons.wikimedia.org/wiki/File:San_Francisco_Chinatown_alley.jpg
-
Red fox at Kodiak National Wildlife Refuge. Lisa Hupp, U.S. Fish and Wildlife Service. Public domain, U.S. government work (PD-USGov-FWS). https://commons.wikimedia.org/wiki/File:Red_fox_(8385104980).jpg
-
Waves at sunset, Rialto Beach. Brent Lawrence, U.S. Fish and Wildlife Service. Public domain, U.S. government work (PD-USGov-FWS). Still from a USFWS video of Rialto Beach. https://commons.wikimedia.org/wiki/File:Rialto_Beach_sunset_(53676352325).jpg
-
Seeded sourdough, Sonoma, California. Sarah Stierch. CC0 1.0 Universal. https://commons.wikimedia.org/wiki/File:Sourdough_bread_-_January_2026_-_Sarah_Stierch_01.jpg
-
Galactic center in infrared, Spitzer Space Telescope. NASA/JPL-Caltech/S. Stolovy (Spitzer Science Center/Caltech). Public domain, U.S. government work (PD-USGov-NASA). https://commons.wikimedia.org/wiki/File:Milky_Way_IR_Spitzer.jpg
-
Laptop and notebook on a wooden desk. Pixel.la Free Stock Photos. CC0 1.0 Universal. https://commons.wikimedia.org/wiki/File:Coffee-desk-notes-workspace_(24243718641).jpg
-
Bumble bee on a blanket flower, Big Stone National Wildlife Refuge. Mike Budd, U.S. Fish and Wildlife Service. Public domain, U.S. government work (PD-USGov-FWS). https://commons.wikimedia.org/wiki/File:Bumble_bee_on_blanket_flower_(54015636001).jpg
-
Red barn at River Road Farm, Millstone Valley, New Jersey. Lardner/Klein Landscape Architects, for the Federal Highway Administration. Public domain, U.S. government work (PD-USGov-DOT). U.S. National Archives NAID 7720407. https://commons.wikimedia.org/wiki/File:Millstone_Valley_Scenic_Byway_-_Red_Barn_of_River_Road_Farm_-_NARA_-_7720407.jpg
.jpg){kind=link}
.jpg){kind=link}
.jpg){kind=link}
.jpg){kind=link}
.jpg){kind=link}
.jpg){kind=link}
Why a tidy picture is not proof
A centered path, a heart in the foam, and a barn with even boards show up often in generated pictures. So does a sky with no dust and no streak. Cameras do all of those things too. The Hoh Trail photo is messy because a log fell there, not because mess is a seal of authenticity. The Spitzer picture is dazzling and still a real NASA exposure. The tells under each reveal are there to practice a close look. They are not a verdict you could use on someone else.
If you want a different kind of practice, the AI or Human writing test asks you to read a passage, the Prompt Battle compares two prompts without calling a model, and the AI detector reads text you paste. The detector and the AI humanizer send text to an AI. This page does not. The Tech Myth or Fact deck stays in the browser too.
FAQ
Does this upload my answers?
No. The quiz runs in your browser. The only thing saved is your best score, in this browser's local storage. The photo you saw and the choice you made are not stored and are not sent anywhere. The page does not call an AI.
Where do the real photos come from?
Each real photo is a public-domain U.S. government image or a CC0 file. The Credits section names the author, the license, and the Wikimedia Commons file page where that license is recorded. The pictures are stored with this page. Playing does not fetch them from those sites.
Were the AI photos made live by a model?
No. The AI photos were generated ahead of time for this game and saved on this site. The reveal labels them Generated by AI for this site. Playing does not send the picture to an API.
How does a round work?
Each round is 10 pictures, five real and five AI, drawn from a deck of 20 and shuffled. You choose Real or AI, then see the source and one or two tells. Play again deals another 10. Your eye accuracy is the share of the round you got right.
Are the tells reliable?
No. Even lighting, repeated shapes, and a too-clean center are habits, not proof. A camera can make a tidy picture, and a model can copy a crooked one. The note under each reveal says so.
What is saved on this device?
Only your best score. Clearing site data for cyberscryb.com removes it. The quiz does not ask for an account, and leaving the page drops the current round.