ZONN.ai Forensic Report

Case · 9C49FB80 · IMAGE

Analyzed byAnonymous
ZONN Analysis
0

Inconclusive

Signals are mixed or weak. We can't tell with confidence — context, source, and your own judgement matter here.

Signal ConfidenceWeak · 19/100

Analysed Specimen

Analysed content
No flagged regions

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.

Multiple detectors unreachable

Note

4 ML detectors did not respond (Ml Img Cifake, Ml Img Deepfake, Ml Img 3class, Ml Python Siglip). The verdict was computed with reduced evidence; reliability is lower than usual.

Just below the AI threshold

Note

The score (56/100) is right at the boundary between inconclusive and AI (60). A single signal flip could swing the verdict — treat this as suggestive rather than definitive.

Uneven compression (ELA)

AI evidence

ELA coefficient of variation is 2.09 — different regions show noticeably different compression levels. Common with composites, edits, or AI generations.

Weak overall confidence

Note

Aggregate verdict confidence is 19/100. Several detectors returned uncertain answers or were offline. Read the verdict as a guideline, not as a final answer.

No ICC color profile

AI evidence

The 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.

Why this verdict

  • Error Level Analysisread real · 0/100

    Error Level Analysis. Re-saves and diffs to expose uneven compression regions.

  • Noise Patternflagged AI · 75/100

    Sensor-noise (PRNU) fingerprint check. Real cameras leave camera-unique noise.

Model Agreement

100%

Variance across 8 ML detectors. Higher agreement means the models converged on the same reading; lower agreement means treat the verdict with care.

Evidence — 18 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 Models8 detectors · mean 50
▸ expand
AI Image Detector
50
Deepfake Detector
50
3-Class Detector
50
SigLIP AI Detector
50
RIGID (DINOv2)
50
Fourier Zero-Shot
50
NPR Artifacts
50
GramNet Texture
50
Pixel & Frequency Forensics7 detectors · mean 42
▸ expand
Error Level Analysis
0
Noise Pattern
75
Pixel Analysis
35
Edge Consistency
35
Color Distribution
39
Frequency Analysis
55
Compression Quality
55
Provenance & Metadata3 detectors · mean 57
▸ expand
ICC Profile
62
Metadata
59
C2PA Provenance
50

Image Quality

Dimensions
1254 × 1254 px
Aspect
1.000
File size
1.16 MB
Bytes / pixel
0.774

Edge Consistency

CV 0.860
Cell 1: 1.7621Cell 2: 3.2418Cell 3: 2.2300Cell 4: 1.6835Cell 5: 2.7261Cell 6: 13.6830Cell 7: 14.5369Cell 8: 3.4642Cell 9: 4.7000Cell 10: 13.2351Cell 11: 13.0891Cell 12: 5.0892Cell 13: 1.7052Cell 14: 3.0099Cell 15: 3.0253Cell 16: 1.6129

Per-region edge density (4 × 4 grid). Uneven distribution may indicate localized editing or splicing; uniform fields are typical of fully synthetic outputs.

Range: 1.612914.5369

Noise Fingerprint

Variance
78.64
Std deviation
8.87
Mean
-0.0
Spatial corr.
1.393
Mean Δ
0.63
σ
1.31
CV
2.090
Uniformity
-1.090

Provenance

Source Dossier

PlatformDirect upload
Author
Content Typeimage
Analyzed OnMay 10, 2026, 9:36 AM
Analyzed byAnonymous