File integrity
Compares the file name, declared format and binary signature, then creates a SHA-256 fingerprint for the exact copy analysed.
KIR Verify · Private file checks
Trust the file less. Verify the evidence more.
A private technical check for images and PDF documents. KIR Verify looks for file inconsistencies, editing traces, recorded AI markers and suspicious structure—and can compare a suspected copy with an original.
Local No uploadExplainable Evidence shownResponsible No false certainty
Try KIR Verify
A clearer first look at a questionable file.
One guest check needs no account or server upload. Sign in for your three-check allowance. The browser scan checks signatures, metadata, AI-workflow markers, image structure and compression behaviour locally; KIR does not receive file contents.
Checking secure access…
What it checks
Several signals. One explainable report.
Each result shows what was observed and why it matters, so reviewers can make a better decision rather than trust a mysterious score.
History & provenance
Looks for editing software, recorded AI-workflow declarations, EXIF/XMP data and Content Credentials markers without treating missing metadata as fraud. It is not a content-level AI classifier.
Structural signals
Reviews PDF revision markers, embedded or active content, trailing data and other features that may deserve closer examination.
Visual consistency
Creates an amplified JPEG recompression map to help a human reviewer notice areas with different compression behaviour.
Built for real workflows
From a suspicious receipt to a document that needs human review.
KIR can tailor verification intake and review workflows for insurance, recruitment, property, education, legal, finance and internal compliance teams.
Receipts & invoicesCertificates & statementsIdentity-document imagesInsurance evidenceProperty photographsRecruitment documents
Responsible use
Request an expert review Technical screening—not a digital lie detector.
KIR Verify cannot prove authenticity, authorship, AI generation, fraud or factual truth. Authentic files can contain indicators, and altered files can pass undetected. Results must be considered alongside the original source, surrounding context and appropriate human expertise.