According to foreign media reports, ByteDance has been secretly using OpenAI's technology to build its own large language model in order to gain an advantage in the generative AI field, which violates OpenAI's terms of service. Currently, ByteDance's account has been suspended by OpenAI.

Foreign media pointed out that this practice is considered "impolite" in the AI industry and also violates OpenAI's terms of service, which explicitly prohibits the use of its models to "develop AI models that compete with our products and services." ByteDance obtained access to OpenAI through its acquisition by Microsoft, but Microsoft has the same regulations.

Internal documents obtained by foreign media reveal that ByteDance relied on OpenAI's application programming interface (API) for every step of developing its large language model codenamed "Project Seed," including training and evaluating the model. Employees involved in "Project Seed" were well aware of the risks associated with this behavior. Their internal communication on the messaging platform Lark showed discussions on how to hide evidence through "data anonymization." Foreign media reported that ByteDance employees excessively used OpenAI's technology to the point where the employees of "Project Seed" frequently reached the access limit of the OpenAI API.

The internal documents also show that ByteDance mainly used OpenAI's technology in the early stages of "Project Seed." Several months ago, the company instructed the team to stop using text generated by GPT at "any stage of model development." Around the same time, the company launched its own AI model called "DouBao" and put it into operation for "Project Seed." However, ByteDance continued to use the API in violation of OpenAI and Microsoft's terms of service, including evaluating the performance of the DouBao model. Someone familiar with the internal situation at ByteDance said, "They say they want to ensure legality, but in reality, they just don't want to get caught."

Yodi Seth, a spokesperson for ByteDance, responded by saying that data generated by GPT was used for model annotation in the early stages of "Project Seed" and has been removed from ByteDance's training data around mid-year. "ByteDance obtained permission from Microsoft to use the GPT API. We use GPT to power products and features in non-Chinese markets, but we use our self-developed model to power DouBao. DouBao is only available in China," Seth said.

Nick Felix, a spokesperson for OpenAI, confirmed that ByteDance's account has been suspended. "All API customers must comply with our usage policies to ensure our technology is used for good. While ByteDance rarely uses our API, we have suspended their account during further investigation. If we find that their usage does not comply with our company policies, we will require them to make necessary changes or terminate their account," Felix said.

Frank Shaw, a spokesperson for Microsoft, stated, "Azure OpenAI services and other Microsoft AI solutions are part of our limited access framework, which means all customers must apply for and obtain Microsoft's approval to access them. We have also established standards and provided resources to help our customers use these technologies responsibly and comply with our terms of service. We have processes in place to detect abusive behavior and stop their access when enterprises violate our code of conduct."