← Back to context

Comment by egypturnash

8 days ago

thanks for making everything that much shittier just so you can steal everyone's data and present it as your own, AI companies!

The tiniest relief is knowing that models will be distilled and "copied" in smaller models of equal capabilities in ~6/12months, since their output can't be copyrighted and will be used to improve others. Kinda ironic

you can't steal something that is in public domain and one which you make readily available by publishing it online because there is no provable cost of damage to you by someone scraping and training their models.

if you really think what you offer has value, put it in behind a paywall and see how many people will consume it then, probably not a lot.