How Teyit Debunked an AI Video of a Turkish Flag in Times Square
A video spread on social media showing a giant Turkish flag hanging over New York's Times Square. It never happened. Turkish fact-checker Teyit traced the clip to AI generation using three independent checks. One of them was UncovAI.
By the UncovAI team · Published 12 November 2026
Key takeaways
- The claim: a video shows a huge Turkish flag hung in Times Square, New York.
- The verdict: false. Teyit classed it as AI-generated content on 11 November 2026.
- The evidence: visible distortions in the flag, no credible news coverage, and a 95% AI-generation likelihood from UncovAI's detection tool.
- The lesson: the strongest fact-checks stack several kinds of evidence. A detector is one of them.
What did the viral video claim?
The video was shared on social media with the claim that it showed a giant Turkish flag unfurled over Times Square. The setting is one of the most filmed places in the world, which is part of why a clip like this looks plausible. Everyone knows the screens. Few people stop to ask whether a specific event actually took place there.
Teyit, a Turkish fact-checking organisation, examined the video and rated the claim false. You can read the full analysis on Teyit's website (in Turkish).
What visual errors gave the AI video away?
Teyit went through the footage second by second. The flag itself broke down in ways a real flag would not.
| Timestamp | What the flag shows |
|---|---|
| 0:07 | Two crescents on the same flag. |
| 0:09 | The star is missing, and the flag still carries two crescents. |
| 0:13 | The crescent and star merge into each other. |
These are typical generative artefacts. A model reproduces the look of a symbol, not its rules. A crescent and star are a fixed design. When a video model has to keep them consistent across a rippling, moving surface, the geometry drifts from frame to frame.
Why did the sourcing check matter?
Visual errors raise suspicion. They do not close the case. A skilled editor can fix the flaws, and a real video can contain compression glitches that look like flaws.
So Teyit also searched credible sources with keywords describing the event. It found no news report of a giant Turkish flag hung in Times Square. For an event of that size, in that location, the absence of coverage is strong evidence on its own.
What did AI detection add?
Teyit then ran the video through UncovAI. The tool returned a 95% likelihood that the footage was produced with AI tools. Our video detection analyses the frames and the audio track separately and returns a verdict with a confidence score, plus a heatmap of the regions that drove the result.
The score is a probabilistic output, not a certificate. Its value here is that it comes from a different kind of evidence than the other two. A human reviewer spotted the flag errors by eye. A search established there was no coverage. The detector looked at pixel-level patterns that people cannot see. When three independent methods point the same way, the conclusion holds up.
A detection score should not be the only basis for a verdict. Teyit used it as one of three converging lines of evidence, which is the right way to use it.
A workflow you can copy
The Teyit analysis maps onto a simple routine for any suspicious clip.
- Scrub through the video. Pause on details that should stay constant, such as flags, logos, text and hands. Watch whether they change between frames.
- Search for the event. Use keywords in credible news sources. A big public event leaves a trail.
- Run a detector. Upload the clip to a tool that gives a verdict and a confidence score. The UncovAI browser extension lets you do this without leaving the page where you found the video.
- Weigh the evidence together. Treat any single signal as a lead. Treat agreement across methods as a finding.
Frequently asked questions
Is the video of a giant Turkish flag in Times Square real?
No. Teyit rated the claim false on 11 November 2026 and classed the video as AI-generated. It found visual distortions in the flag, no credible news report of the event, and a 95% AI-generation likelihood from UncovAI.
What visual errors gave the video away?
Two crescents on the flag at 7 seconds, no star at 9 seconds, and a crescent and star merging into each other at 13 seconds.
What does a 95% AI detection result mean?
It is the detection model's confidence that the video was generated by AI. It is a probabilistic output, not proof on its own. In this case it was one of several independent lines of evidence.
How can I check whether a viral video is AI-generated?
Combine methods. Look frame by frame for distortions, search credible sources for coverage of the claimed event, and run the video through an AI detection tool. Our video detection page explains what the tool reports.
Check a video yourself
Upload a suspicious clip and get a verdict, a confidence score and a heatmap in seconds.
Get Started Free →