ZONN.ai Forensic Report
Case · 7205991A · IMAGE
Probably AI-generated
More signals lean toward AI generation than real, but some give weaker readings. Treat with caution.
Analysed Specimen
Heads up — 5 things to know
Why this analysis might be off
We highlight every disagreement and unusual signal we found so you can judge for yourself. Stronger warnings come first; informational notes are at the bottom.
ML detectors disagree with each other
Note2 models confidently say "AI" (Itsnotai V2, Xrayon Convnext) while 3 confidently say "real" (Ml Python Siglip, Ml Commfor, Bombek1). This image sits at the edge of what ML can decide — manual review is recommended.
No metadata at all
AI evidenceThe file has no EXIF, XMP, IPTC, or ICC metadata. This is common for social-media re-uploads and for many generator outputs — real camera files almost always carry an ICC profile.
Upsampling artifacts in the frequency domain
AI evidenceFFT analysis found strong upsampling patterns — a fingerprint of diffusion-model VAE decoders (latent → pixel-space upscale).
ML detectors see this image differently
NoteML scores span a wide range (5–100). Different architectures read different feature spaces; the majority vote strengthens the consensus, but no single model is fully reliable here.
No ICC color profile
AI evidenceThe image does not embed an ICC color profile. Real camera files almost always carry sRGB or Adobe RGB profiles — a missing profile is often a sign of generator output or re-encoded media.
Origin Check
Trace this image elsewhere
Cross-reference the source against major reverse-image services. Each link opens in a new tab with the image URL preloaded — ZONN.ai does not re-upload the image.
Google Lens
Yandex
TinEye
Bing VisualWhy this verdict
- INA v2 (FLUX/MJ)flagged AI · 100/100
BEiT-Large dual-head classifier trained on FLUX, Midjourney, and real photo corpora.
- xRayon ConvNeXtV2flagged AI · 97/100
ConvNeXtV2 detector trained on FLUX, DALL-E 3, SDXL, SD3.5, and Midjourney v6.
Model Agreement
Variance across 6 ML detectors. Higher agreement means the models converged on the same reading; lower agreement means treat the verdict with care.
Evidence — 16 detectors reviewed
What each detector saw
Each detector independently gave this imagea score from 0 (definitely real) to 100 (definitely AI). The score above is their weighted consensus — detectors with higher confidence count more. No single detector decides; you read the spread.
ML Models6 detectors · mean 48▸ expand▾ collapse
Pixel & Frequency Forensics7 detectors · mean 51▸ expand▾ collapse
Provenance & Metadata3 detectors · mean 54▸ expand▾ collapse
Image Quality
- Dimensions
- 960 × 1286 px
- Aspect
- 0.747
- File size
- 2.31 MB
- Bytes / pixel
- 1.963
Frequency Analysis
Edge Consistency
Per-region edge density (4 × 4 grid). Uneven distribution may indicate localized editing or splicing; uniform fields are typical of fully synthetic outputs.
Range: 2.0698 – 19.6980
Noise Fingerprint
- Variance
- 76.56
- Std deviation
- 8.75
- Mean
- -0.0
- Spatial corr.
- 2.607
- Mean Δ
- 1.59
- σ
- 1.68
- CV
- 1.059
- Uniformity
- -0.059