Earlier quoted context omitted.
There is a large disclaimer that states, among other things, "We strive to ensure accuracy and quality using authoritative sources and AI-based validation; however, we make no guarantees regarding completeness, accuracy, or timeliness. Always confirm nutritional data independently when accuracy is critical." on every page on the website where that kind of in-depth data is available.
At that point, if you are not sure a data point is accurate, should you really display it ? You have no proof appart from "The LLM said it was ok" which is kind of poor.
My guess is that this dataset is probably more accurate on the whole than many datasets used by the kinds of calorie-tracking apps that outsource their collection of nutrition information to users. But an analysis would be required.
Regardless, the only workable approach is to describe the provenance of your data and explain what steps have been taken to ensure accuracy. Then anyone who wants to use the data can account for that information.