I worked with a company that used product data from competitors (you can debate the morals of it, but everyone is doing it). Their crawlers were set up so that each new line of requests came from a new IP.. I don’t recall the name of the service, and it was not that many unique IP’s but it did allow their crawlers to live unhindered..

They didn’t do IP banning for the same reasoning, but they did notice one of their competitors did not alter their IP when scraping them. If they had malicious intend, they could have changed data around for that IP only. Eg. increasing the prices, or decreasing the prices so they had bad data..

I’d imagine companies like OpenAI has many times the IP, and they’d be able to do something similarly.. meaning if you try’n ban IP’s, you might hit real users as well.. which would be unfortunate.