OsintNET Image Authenticity Scanner helps analysts investigate suspicious images with pixel-level AI detection, metadata forensics, generator traces, compression clues and clear evidence cards.
Choosing an image starts all checks automatically and sends an analysis image to Sightengine. Files over 4 MB use a resized JPEG copy; original signatures are checked locally. The automatic JPEG comparison tests export sensitivity, not accuracy.
Original files can contain camera, lens, timestamp, GPS and editing software clues. OsintNET surfaces those signals and explains their limits.
Structured PNG workflow and software fields can identify a declared generation workflow. These unsigned fields can be edited or forged; filenames and arbitrary binary text do not become AI evidence.
The C2PA verifier checks embedded signatures and content binding against a pinned trust list. Signed AI declarations are distinguished from untrusted or invalid manifests. Signed history does not prove that depicted events happened.
Compression patterns, residual noise, edge density and clipped exposure regions help build a review profile, especially when metadata has been stripped.
Türkçe kullanım için araç, bir görsel yapay zeka mı değil mi sorusuna kanıt tabanlı cevap üretir ve sonucu raporlanabilir bulgulara ayırır.
Every result is presented with confidence, source context and export options so analysts can document why an image was flagged.
OsintNET offers automatic Sightengine pixel analysis, local C2PA signature verification, structured metadata and visual forensic maps. The score is an uncalibrated model response, not a probability of authenticity. Missing model results remain undetermined.
Yapay zeka görsel tespiti için görselin piksellerindeki üretim izleri, EXIF ve XMP metadata, dosya yapısı, sıkıştırma davranışı ve varsa üretici araç izleri birlikte değerlendirilir.
No. Social networks and messaging apps often strip EXIF metadata. Its absence is not AI evidence and never changes the detector score.
Start with the original file, run AI detection, inspect metadata, compare older web copies with reverse image search and document every finding with confidence and source context.
Pick the module that matches your target and keep each clue connected to its source, confidence and investigation context.