Artificial intelligence (AI) models are becoming increasingly complex and have amazing capabilities, but the training process is also increasing. ChatGPT is still the AI with the highest number of users but is gradually being overtaken by competitors such as Gemini, Claude and DeepSeek. In this situation of peacock position being threatened, OpenAI sent a memo to the US House of Representatives Select Committee and accused DeepSeek of stealing American AI models to train their own models.
DeepSeek is accused of using a distillation technique that asks large and mature models to evaluate the output of a new model. This distillation technique will transfer the knowledge of the American AI model to the DeepSeek model. In addition to the distillation technique, DeepSeek is also accused of various ways to avoid their access being blocked by OpenAI.
This technique will allow DeepSeek to drive model development faster and avoid costs. Reactions on the internet are generally laughing at OpenAI because it is no secret that they use various intellectual works in the form of text, images, videos and audio to train their models without the owner's permission.
