Fondsites Diagnostics
Reality Check
AI Image Reality Check
Use a cautious workflow for visual clues, source tracing, metadata, and context without pretending visual AI detection is certain.
Image verification is a workflow, not a visual guessing game. AI generation, editing, old-image recaptioning, missing source context, and fake accounts can look similar.
This page does not produce a certainty score. It helps you slow down, preserve context, and move toward stronger evidence.
The strongest verification signal is usually provenance, not pixels. Reverse-image search, the earliest findable posting, and whether a credible outlet has independently carried the image will resolve more cases than staring at hands and shadows, because generation artifacts change with every model release while sourcing discipline does not. When the stakes are real — money, reputation, safety — the honest output of this workflow is often “unverified,” and acting on unverified is a choice you make knowingly.
Interactive Diagnostic
AI Image Reality Check
Select the closest symptom and context. The output is a cautious first test, not a certain diagnosis.
Diagnostic output
Choose the closest symptom to see likely causes and a first reversible test.
Symptom, Cause, Action Table
| Symptom | Possible cause | First action |
|---|---|---|
| Visual artifacts | AI generation, editing, compression, or low-quality repost. | Use source checks before certainty. |
| Dramatic caption | Old image, wrong location, or missing context. | Reverse search and date-check. |
| Urgent account | Impersonation or scam pressure. | Verify through known channels. |
| No source | Repost chain or deliberate ambiguity. | Ask for origin and preserve the link. |
Fast checks
- Separate the image from the claim made about it.
- Search for earlier copies and different captions.
- Check account history and known-channel confirmation when people or money are involved.
- Preserve screenshots and URLs before confronting or reporting.
What not to change yet
- Do not declare an image fake from visual clues alone.
- Do not upload sensitive private images to random detector sites.
- Do not engage with urgent money or identity requests before known-channel verification.
Assumptions and limitations
Visual AI detection is not reliable enough for certainty.
Metadata may be stripped, altered, or absent; provenance signals are useful but not universal.