Editorial policy

Public Health Trace doesn't have a newsroom or editorial staff writing original stories. This page explains exactly how the titles, summaries, and text you see on the site are produced, so it's clear what is and isn't editorial judgment.

How titles are generated

Recall titles are chosen deterministically from the official fields a source provides, in a fixed order: the recalled product's brand name, if published; otherwise the recalling company's name; otherwise the product description from the recall record; and only as a last resort, the source's own recall number or notice identifier. No title is written by a person or a language model -- the same input always produces the same title.

No AI-generated content

Public Health Trace does not use AI to write summaries, rewrite source text, or generate narrative copy anywhere on the site. Every recall and outbreak record shown is built directly from fields published by the source agency (the CDC, FDA, or USDA), formatted for readability -- never rewritten, embellished, or interpreted.

Source attribution

Every recall, outbreak, and surveillance record links back to the official source it came from. When Public Health Trace shows a field the source didn't publish (for example, lot codes not captured by a given importer -- see methodology), it says so explicitly rather than leaving a misleading blank or inventing a value.

Corrections and review

If a record looks wrong, our corrections policy explains how to report it and what happens next. Because Public Health Trace republishes agency data rather than authoring it, most corrections trace back to either an import error on our side (which we fix directly) or an error in the original source record (which we note and point you to the agency to correct at the source).