How FridgeFox checks product and content claims.

How are product claims tested?

Product claims are checked against the current application route, visible interface, configuration contract, and automated tests. Scan, pantry, recipe, authentication, billing, consent, and indexing behavior are tested separately so one successful flow does not imply another.

How is AI scanner behavior evaluated?

A scanner evaluation must define its sample, image conditions, expected labels, model and application version, review rules, and error categories before results are collected. Packaging, occlusion, lighting, image quality, and category ambiguity are reported as limitations rather than hidden.

How are food and storage claims sourced?

Food-safety and storage statements use current government, public-health, university, peer-reviewed, or recognized international sources where possible. FridgeFox does not turn a general source into a guarantee about a specific item.

What is not treated as evidence?

• Generated placeholder results or synthetic testimonials.
• A small internal example presented as a population-wide accuracy rate.
• A client-side payment success message presented as proof of purchase.
• An AI suggestion presented as food-safety, allergy, nutrition, or medical advice.

How are changes and corrections handled?

Published and updated dates change when a meaningful claim, method, result, or product behavior changes. See the corrections process, editorial policy, and research section.