ByteDance’s use of OpenAI’s technology to develop competing products is frowned upon in the artificial intelligence (AI) industry, The Verge reported. Not only that, it also violates the terms of service of the company behind ChatGPT.
Under the terms, OpenAI customers are prohibited from “developing any AI models that compete with our products and services.” They are also prohibited from using “any means to extract data from the service other than as permitted through the API.” The API is an application programming interface that allows developers to use GPT to create their own applications.
ByteDance knew this, but continued to use the API to train and compare its Project Seed model. The Verge said it had seen internal ByteDance communications that instruct employees to hide evidence using a technique called “data desensitization.” This is the process of editing or removing sensitive information from data to protect it from unauthorized access or discovery.
ByteDance then asked employees to stop using the API to develop Project Seed when the Doubao chatbot was licensed for use in China. However, The Verge reported that the API was still used to evaluate the company's chatbot performance.
OpenAI confirmed on December 15 that it had suspended ByteDance’s account for violating its terms of service. Spokesperson Niko Felix said all API customers must adhere to the company’s usage policy to ensure the technology is used for good.
Even though ByteDance only used OpenAI’s API at a minimal level, they were suspended while a further investigation was conducted. If non-compliance is found, OpenAI will request that ByteDance make the necessary changes or terminate the account.
In a statement to Business Insider, ByteDance denied any wrongdoing and said it had a license to use GPT's API. It uses GPT for products and features in markets outside of China and uses a self-developed model for Doubao, a chatbot that is only available in mainland China.
(According to Insider)
Source
Comment (0)