Can you spot the fake?
A day at the Hogeschool Utrecht expo, told short.

The morning started with a poster, a roll of sticker dots and a very simple idea. If people believe a face is real, they should be able to say why. So we put up a board that said "This is fake news!" and, next to it, another that asked the only question that matters: can you spot the fake?

Students and staff slowed down as they walked past. Some laughed at the big red words. Others stopped and looked closer. The rules were easy. Look at the pictures, decide which one is made by a machine and put your sticker under the face you think is fake.

The dots told the story before we did. Some faces collected stickers fast, and a few real photos got more doubt than the fakes. Nobody was embarrassed about it. That was the point. If a room of bright people can be fooled, a scrolling feed at midnight does not stand a chance.
The moment someone says "wait, how do I know?" is the moment the tool starts to matter.
We also learned something about surveys. Most visitors were not keen to fill one in, a form looks like homework. But the moment we put the app in their hands and showed it working, their interest changed. They wanted to try it, then they were happy to tell us what they thought. A demo beats a questionnaire every time.
That question led to the phone. We opened the Android app and showed Spot the Fake on a real screen, then Claim Check, then Election Watch with the Dutch fact-checks. People checked a claim they had actually seen that week, and that is when the conversations got long. A few asked about the browser extension. A few asked how we decide what is trustworthy. We pointed them to the independent fact-checkers whose work we show, because we do not want to be the judge. We want to make the real verdicts easy to find.
By the afternoon our voice was tired and our notebook was full. Feedback came in about thumbnails, about language, about what students would use on a bus ride. A good part of it already went into version 0.1.6 of the Android app.

Thank you to everyone who stopped, voted, argued and asked hard questions, and to Hogeschool Utrecht for hosting us. If you missed it, the app is open for testing. Come and try to beat the fakes yourself.
What the survey told us
26 people filled in our expo survey between August and the expo day. Here is what stood out.
72%
of the people who played the image challenge (18 of 25) picked the AI image as the real one.
8.2 / 10
average score for how important it is to know if content is real.
12 of 23
said they would definitely or probably use a tool like this every day.
Mixed crowd
9 researchers, 4 students, 4 teachers, 4 community leaders and 3 journalists.
People told us they check content by searching for other sources, asking a friend, doing a reverse image search or zooming in. Almost nobody had a tool for it. Ease of use scored 3.8 out of 5, and 10 of 23 people said a tool like this should simply be free, which is why the core stays free.
"Develop one for handheld devices."
That request from a community leader is why the Android app exists. Another visitor asked when a page can be called credible, because every site they tested scored between 50 and 60 percent. We take that as a reminder to explain the score better, so see how it works on the Transparency Score page.
More photos






Watch the clips
Ready to check what you read?
Install free, join the Android test or follow the WhatsApp channel for new releases.