I have no problem with using publicly accessible data to train AI models, as long as that model is also publicly accessible. (See llama, mistral, etc.) But what OpenAI is doing is wrong because they’re taking that public data and hoarding it for profits instead of for public benefit.
Telling people copyright shouldn’t apply to them while hiding the fact that copyrighted material was used to train it seems quite hypocritical.
I have no problem with using publicly accessible data to train AI models, as long as that model is also publicly accessible. (See llama, mistral, etc.) But what OpenAI is doing is wrong because they’re taking that public data and hoarding it for profits instead of for public benefit.