[cuttlefisch]

If it is publicly accessible, they cannot limit the means by which you access the information from the perspective of the law. This was seen in the case of Oracle & another company scraping their technical documentation IIRC (https://www.eff.org/deeplinks/2018/01/ninth-circuit-doubles-...). The site may take action to prevent this access, etc. but they cannot pursue legal action for this alone (unless this has changed in the last couple of years). There was a similar precedent with Linkedin maybe 5-7 years ago.

That being said, depending on the target they can make things difficult if they know you're doing it. In general, scraping data itself, then transforming that data for use is reasonably safe, but using the scraped content in unprocessed form can be problematic. Selling the collected data to users without processing it sounds like it could cause problems both from the target companies, as well as via the customer's perception of how you acquire the data. Processing the data to show something like variations in recipes per region, categorizing different recipes into styles based on ingredients, cook time, complexity, etc. are all value-adds which make your data more useful than the raw data-set, and make a stronger argument for the sale of your dataset.

Of course IANAL, and I welcome anyone else to add to or contradict this info.