Maybe Depseek Distilling’s Models to train yours, and perhaps that is a violation of the operating terms that OpenAi has published. But “extracting information and putting it into use” feels like a fair description of what Deepseek has done here. If Depseek’s work really wasn’t possible without the work Openai had already done, maybe Deepseek should think of compensating OpenAi in some way?
This type of hypocrisy makes it difficult for me It is worth having a conversation about beforehand.
A last comment of irony in Andreessen Horowitz’s comment: there is some Husming about the impact of a copyright infringement decision on competition. Having to license works with copyright on scale “would be invested for the benefit of the largest technological companies, those with the deepest pockets and the greatest incentive to maintain the models of closed to the competition.”
“A multimillion -dollar company could afford to license training data with copyright, but new smaller and agile companies will be completely excluded from the development career,” continues the comment. “The result will be much less competition, much less innovation and most likely the loss of the position of the United States as a leader in the global development of AI.”
Part of the agita on Depseek of the industry is probably wrapped in the last bit of that statement, which apparently a Chinese company has defeated an American company with something. Andreessen himself referred to the Deepseek model as a “Sputnik moment“For the AI ​​business, which implies that American companies need to catch time that restricts similar access to your own working.
Good luck with that!