Colpo per OpenAI in Germania perché il tribunale stabilisce che i testi delle canzoni siano usati illegalmente
https://www.yahoo.com/news/articles/blow-openai-germany-court-rules-151638208.html
di MetaKnowing
Colpo per OpenAI in Germania perché il tribunale stabilisce che i testi delle canzoni siano usati illegalmente
https://www.yahoo.com/news/articles/blow-openai-germany-court-rules-151638208.html
di MetaKnowing
9 commenti
Good. More of this please.
Awesome! Fuck OpenAI’s theft machine.
Good.
yeah, data theft is a serious crime, especially now that they have turned their crime into money and want to make billions or trillions unless this bubble finally pops
stop AI and make them abide the law for once
Ok, now if you remember some song lirycs, you must pay a fee, because you are storing someone’s data in your brain.
What a wonderful world Anti-AIs build for themselves.
> The court concluded that the song lyrics, used to train ChatGPT, had been stored in OpenAI’s systems. The AI firm had argued in vain that the system only learns from, but does not store, such material.
Really, how did this court PROVE that AI storing exact data in their systems? Did they make an audit in OpenAI? Did they even know how AI works? Without “scraping copyright machine” nonsense?
They are right as AI companies should be using public domain data for training. They learn patterns in the data to reproduce it. LLM training literally penalizes the model for predicting the next token in the lyrics incorrectly.
It’s not like they are secretive about stealing stuff. It’s just that they hope it’s OK.
i fail to understand this completely: “i sing this song loudly but you’re not allowed to remember it, I’ll show you this painting but you’re not allowed to paint another like me”
Good. AI companies should have to pay for the IP they use for training.