OpenAI says DeepSeek may have used their data for their model

OpenAI is saying that DeepSeek may have ‘inappropriately’ taken data from their model to develop their own AI chatbot. DeepSeek has been on a crazy run this week, shaking up the AI arms race and the stock market.

As per an OpenAI spokesperson via NBC, DeepSeek may have used distillation to develop their model. For context, Distillation is a training method used to train a large language model (LLM) with data generated from another LLM.

The spokesperson said that OpenAI is aware and reviewing indications where DeepSeek may have inappropriately distilled their models. The company is bound to share information once they know more.

Distillation is common for the industry, but is often prohibited from LLM terms of service. The company’s spokesperson added that OpenAI is working closely with the US government to protect their technology.

Weirdly enough, OpenAI’s ChatGPT was also accused of inappropriately using content it’s restricted from. Of course, this was subject to multiple lawsuits.

DeepSeek has yet to respond to the allegations made at the time of writing this article. We’ll keep readers updated on the matter as this is still a developing story.