Dana Fried on Nostr: OpenAI complaining that NYT unfairly "tricked" ChatGPT into regurgitating an entire ...
OpenAI complaining that NYT unfairly "tricked" ChatGPT into regurgitating an entire NYT article misses the point.
Yes, it's bad that they've copied a bunch of copyrighted material to train their models. No, "Maybe we got it from a third-party site that also pirated the article?" isn't a defense.
But most concerningly, GPT's ability to regurgitate entire articles means that the GPT model being distributed by OpenAI contains and distributes copyrighted work its creators don't have rights to.
Yes, it's bad that they've copied a bunch of copyrighted material to train their models. No, "Maybe we got it from a third-party site that also pirated the article?" isn't a defense.
But most concerningly, GPT's ability to regurgitate entire articles means that the GPT model being distributed by OpenAI contains and distributes copyrighted work its creators don't have rights to.