What is Authenticity AI?
Authenticity AI analyses the photos you upload and cross-references them against known markers of genuine shirts — returning a verdict, a confidence score, and a marker-by-marker breakdown you can inspect yourself.
Authenticity AI is experimental — never use the verdict as the sole basis for a purchase. Always cross-check seller history and provenance.
What it checks
- Physical construction — stitching pattern, badge attachment, jock tag placement, and fabric texture are compared against authentic references.
- Label & tag analysis — font, layout, product code format, and wash-care symbols on the inner label are inspected for consistency with genuine items.
- Confidence score — a percentage that reflects how much evidence was available. The score is capped honestly when label photos are missing or unclear.
- 5-point verdict — every shirt receives one of five verdicts — Verified, Likely Authentic, Uncertain, Suspect, or Counterfeit — based on the totality of the evidence.
How it works
Your photos are sent to a Vision AI model with a structured checklist prompt. The model evaluates each marker, assigns a confidence level, and returns both the individual markers and an overall verdict. The full reasoning is surfaced in the app so you can challenge or override any finding.
Limitations
Authenticity AI works from photos alone. It cannot examine fabric weight, printing technique, or provenance documents. Low-resolution images, unusual angles, or missing label shots will reduce confidence. For high-value items, always consult a specialist authenticator.
Authenticity AI is a beta feature and may produce errors. Your challenges and corrections help us improve it. Have feedback?