AI Disclaimer
What Verifyr’s AI is and what it isn’t. Read this before you act on a verdict.
1. What it is
Verifyr uses computer-vision and language models to estimate the wear of common household items and to recommend a verdict — keep, repair, or replace — together with a five-dimension score and (on paid plans) a short diagnostic reel.
Reels are AI-generated illustrations, not raw inspection footage. Where useful, we annotate them with an “AI generated” marker.
2. What it is not
Not a product certification. Verifyr does not issue conformity-assessment, CE, UKCA, FCC, or any other regulatory marks.
Not a recall lookup. Always check the official recall database for your region; we do not maintain or claim one.
Not an insurance appraisal or forensic test. Verifyr cannot be used as standalone evidence of value, fault, or product liability.
Not an extended warranty or an obligation by Verifyr to repair or replace the items we score.
Not an identity verification or identity-protection service. The brand name "Verifyr" refers to product inspection. We do not verify people or documents.
Not a substitute for licensed professionals. Safety-critical decisions (child seats, helmets, climbing gear, emergency equipment, medical devices) must be made with input from a certified inspector or the manufacturer.
3. Content moderation & safety
Reels and image visualisations Verifyr generates are screened by automated safety filters before they are delivered. Inputs you upload are also screened: we refuse to process imagery that depicts people in private settings, minors, sexual content, or that appears to be the property of someone other than you.
If you spot any visualisation you believe was generated in breach of this policy, email support@mjdtestingservices.com and we will pull it from your account within one business day.
4. How to read a verdict
A verdict is a structured second opinion. The score breakdown shows where the AI sees the largest risk; the rationale paragraph explains why.
Disagree? You should. The final call is yours. If a verdict feels wrong, write to support and we’ll review the reasoning with you.
5. Errors, limits, and bias
AI models can be confidently wrong. They can mis-detect materials in low light, under unusual angles, or for unfamiliar product categories.
We test the model on a broad, anonymised dataset, but cannot guarantee accuracy for every brand, region, or vintage.
6. Contact
Concerns or feedback: support@mjdtestingservices.com.
Effective date: 2026-05-19.