In another way, however, this decision is a cachet of legitimacy on an AI model which caused dismay for Openai during last week. The controversy focuses mainly on the question of whether Deepseek used OpenAi models to produce results (synthetic data) that Deepseek then used to form or refine his own models, a practice often called “distillation”, which is against the conditions of use of Openai.
Since the launch of Deepseek V3 (a large language model that has served as R1 offspring), users have reported that the model is often called Chatgpt, which suggests that at least certain data produced by Chatgpt have been used to refine V3’s behavior. It would not be the first time that IA researchers have come from Openai: IA experts accused Elon Musk’s XAI of doing something similar with his Grok AI model in December 2023.
And this is not the only problem at hand. In addition to the accusation of the terms of service and tweets of OpenAi employees, Microsoft also launched a investigation into Deepseek after Microsoft security researchers discovered that the Chinese company could have extracted substantial data from Fine training via the Openai API during the fall of 2024, according to Bloomberg.
Despite the controversies, the CEO of Openai, Sam Altman, praised the additional Deepseek competition earlier this week. Monday, Altman tweeted: “Deepseek R1 is an impressive model, especially around what they are able to deliver for the price. We will obviously provide much better models and it is also legitimate invigorating to have a new competitor !
In response to R1 climb, Openai should release O3-Mini via Chatgpt from the day later.
