A bombshell report alleges that TikTok’s parent company, ByteDance, secretly utilized technology from OpenAI to build a competing large language model, a direct violation of the AI leader’s terms of service. The project, internally codenamed “Project Seed,” reportedly leveraged OpenAI’s application programming interface (API) for an extended period to train and evaluate its own foundational models.
According to internal documents and sources cited in the report, ByteDance engineers used output data from OpenAI’s models as a benchmark to measure the performance of their own AI. This practice, often referred to as “model stealing,” is explicitly forbidden by OpenAI, as its terms state that API outputs cannot be used to “develop any artificial intelligence models that compete with our products and services.” The data was allegedly used to accelerate the development of ByteDance’s AI, which is crucial for its ambitions to challenge global tech giants in the generative AI space.
Upon being notified of the violation, OpenAI confirmed it suspended ByteDance’s account. In a statement, OpenAI emphasized its policy, noting, “All API customers must adhere to our usage policies to ensure our technology is used for good. Our policy prohibits using our outputs to develop competing models.”
The revelation poses a significant setback for ByteDance, which has been investing heavily to reduce its reliance on third-party AI providers and develop its own powerful models. The incident raises serious questions about the ethical guardrails and competitive practices within the fiercely competitive AI industry. It underscores the challenge of enforcing terms of service when the technology itself can be used to replicate and build rival systems. The fallout could include further legal scrutiny and reputational damage for ByteDance as it navigates its complex relationship with Western tech companies and regulators.
English 
Deutsch