I did some hobby work with a commoncrawl dataset. A year or so later I got contacted by a startup to see if I could do something similar for them. They basically wanted to collect phone numbers and email addresses of businesses from the couple of TB of data. I gave them some disclaimer like: you can use this for good and bad things, I trust it's good and also it is your responsibility to comply with privacy regulations. They ran my code in their own AWS account and probably collected millions of addresses. I think the startup didn't make it. No idea what they did with it but I still feel a bit bad.