Given that big tech has been scraping everything ever written to train LLMs, are... | Hacker News

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

		phtrivier 4 months ago \| parent \| context \| favorite \| on: Anna's Archive: An Update from the Team Given that big tech has been scraping everything ever written to train LLMs, are there specialized prompts to trick models into spitting out copyrighted works ?

baq 4 months ago | [–]

Foundation LLMs are lossy compressed databases, you might never get the exact work back from it.

Nemo_bis 4 months ago | [–]

Yes, if you believe the New York Times: https://www.techdirt.com/2023/12/28/the-ny-times-lawsuit-aga...

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