Does the caption really match the image?
Drop an image or a video, add the caption that goes with it: we compare the two and tell you within seconds whether they really tell the same story — along with the areas of the image that mattered for the verdict. Ideal for spotting an image taken out of context or a misleading caption.
Do the image and the caption tell the same story?
Two posts where the caption fits the image, two where it uses it out of context. The module compares the actual content of the image with the text that goes with it.
Images: Unsplash. Fictional captions, for illustration only.
Do this ad's claims hold up?
Drop a cosmetics ad: we pull out every claim — about ingredients, effects, labels — and we check whether it is accurate or misleading against the EU rules on cosmetic claims. You get a clear verdict, claim by claim, and an overall reliability score.
Is this image trying to manipulate you?
Drop an image or a video: we spot staging and emotional-persuasion techniques — biased framing, dramatization, false urgency signals, alarmist text — often used in advertising, propaganda and viral content. You get a verdict, the elements spotted, and the areas of the image that mattered.
From “designed to make you react” to neutral
Left to right: from the most visually persuasive to the most neutral. The module scores each of these levers — framing, gaze, dramatization — to place a visual on this scale.
Strong persuasion
Strong persuasion
Light stagingNeutralExamples: posters — public domain (Wikimedia Commons); photos — Unsplash. Shown to illustrate the techniques, with no endorsement.
Has this photo been retouched?
Drop a photo: we look for traces of modification — added elements, duplicated areas, photo composites, cosmetic retouching — and we pinpoint the affected areas directly on the image. You get a verdict, the most likely type of modification and a confidence score.
Four photos. Only one has been faked.
This is the kind of material the module examines. The photo in the red frame is a famous historical composite — the other three are authentic. The module pinpoints exactly the suspect area.
Composite
AuthenticAuthentic
AuthenticExamples: Lincoln/Calhoun composite — public domain (Wikimedia Commons); photos — Wikimedia Commons / Unsplash.
Is this image — or this video — authentic?
Three checks to choose from below: a deepfaked face in a photo, a deepfaked face in a video, or an image entirely fabricated by artificial intelligence. You get a verdict, a confidence score, the areas that led to this conclusion and a plain-language explanation.
Can you tell the real from the AI?
Four images: two fabricated by artificial intelligence, two authentic. Hover (or tab) over a card to reveal the verdict — this is exactly the kind of signal the module isolates.
Examples: AI images — Wikimedia Commons; real photos — Unsplash.
Coming soon
This analysis is coming soon.
In the meantime, try “Image & caption”, “Visual manipulation”, “Retouching & editing” or “Advertising & cosmetics”.