So training AI on copyrighted data isn’t a problem unless it spits out the data ... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		wiseowise 8 months ago \| parent \| context \| favorite \| on: Why training AI can't be IP theft So training AI on copyrighted data isn’t a problem unless it spits out the data verbatim. Correct?

lelanthran 8 months ago [–]

Close.

It isn't a problem until it spits out a similar enough copy.

Copyright violation doesn't have to be verbatim; taking a 4k movie and reencoding it to 320x200 before distribution isn't legal.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact