David Sacks has only a few hours after DeepSeek uses Openai models to survey the use of DeepSeek’s application programming interface (API). I was I am reporting.
According to security researchers working in Microsoft, Chinese companies behind the R1 Reasoning model may have excluded a large amount of data using Openai API in the fall of 2024. 。
Anyone can sign up and access the Openai API, but the company’s Terms of Use stipulates that the output cannot be trained.
The company wrote in the Terms of Use, saying that it is prohibited to use output to develop competing models with Openai. In addition, the company states that “automatically or programming (extracted) data or output” is not possible.
The problem of the core seems to be caused by the distillation of AI model developers, which is used to extract from another model using the behavior of teachers and students. It is interesting to see if DeepSeek has found an innovative way to avoid Openai’s rate restrictions and make a large -scale inquiry of the API. If so, there is certainly a legal impact.