ByteDance has confirmed that its use of OpenAI technology to improve its artificial intelligence models was in compliance with the terms of service of the American company, refuting claims that it used the technology to create a competing service.
The parent company of TikTok clarified that a report claiming it secretly used OpenAI technology to develop its well-known large language model called Project Seed, and attempted to conceal evidence by using data anonymization techniques.
ByteDance admitted that a small group of its engineers used the OpenAI API in an unofficial experimental model when the company started exploring large language models earlier this year.
The company explained that this early work ceased in April after submitting a protocol to ensure compliance with OpenAI’s terms of use, which prohibit using GPT product outputs to develop competing models for OpenAI.
ByteDance’s engineering team still relies on the OpenAI API with very limited use of other external models while conducting evaluations and tests, such as result measurement.
The company stated: “ByteDance obtained permission to use the OpenAI API, and is heavily focused on complying with OpenAI’s terms of use,” while OpenAI suspended ByteDance’s access to its service.
The artificial intelligence operator, supported by Microsoft, announced in its statement: “We have suspended ByteDance’s account pending further investigations, although their use of our API was limited. If we find that their use does not align with these policies, we will request them to make necessary changes or close their account.”
OpenAI officially prohibits providing its services in China, where the country is witnessing intense competition with the United States for global dominance in artificial intelligence.
ByteDance’s alleged exploitation of OpenAI technology to build its model comes after platform 01.AI was accused of replicating Meta’s large language model architecture called Llama in its Yi-34B model, as reported by the Hugging Face open-source community.
This claim has sparked controversy, raising doubts about the efficiency of emerging Chinese companies in the field of artificial intelligence.
