TikTok's parent company, ByteDance, was found to have secretly copied ChatGPT, and OpenAI quickly banned ByteDance accounts
by
ByteDance, a major Chinese tech company, previously became a global leader in AI with its TikTok algorithm that finds videos that perfectly match users' preferences, but has fallen behind in the field of generative AI due to the rise of OpenAI. ByteDance, which is trying to make a comeback with its proprietary generative AI 'Doubao,' has violated OpenAI's terms of service by using its APIs to develop a large-scale language model (LLM) under the name 'Project Seed,' according to internal documents revealed by The Verge, a tech news site.
ByteDance is secretly using OpenAI's tech to build a competitor - The Verge
https://www.theverge.com/2023/12/15/24003151/bytedance-china-openai-microsoft-competitor-llm
According to internal ByteDance documents reviewed by The Verge, ByteDance used OpenAI's APIs at almost every stage of development, including model training and evaluation, to develop a foundational LLM codenamed 'Project Seed.'
Such behavior is considered taboo in the AI industry and clearly violates OpenAI's terms of service and Microsoft's terms of service, to which ByteDance directly purchased access, The Verge pointed out.
The terms of use in question are OpenAI's '
The Verge found that employees were discussing ways to 'whitewash' evidence through 'data desensitization' on ByteDance's internal tool, Lark, and that abuse was so rampant that Project Seed employees regularly exceeded API access limits.
While ByteDance boldly used APIs in the early stages of Project Seed's development, by late 2023 it had strictly instructed its development team not to use GPT-generated text at any stage of model development. It was around this time that the company received regulatory approval to release Project Seed in China through its own chatbot, called Doubao.
However, ByteDance continues to use OpenAI's APIs in ways that violate its terms of service, such as evaluating the performance of models for Doubao. A source with direct knowledge of ByteDance's internal affairs told The Verge, 'They say they want to make sure everything is legal, but in reality they just don't want to get caught.'
Meanwhile, when ByteDance was confronted with internal documents by The Verge, the company explained that 'GPT-generated data was used to annotate the Project Seed model in early development and was removed from ByteDance's training data around mid-2023.' In a statement, company spokesperson Jodie Seth said, 'ByteDance licensed GPT's API from Microsoft. While we use GPT to enhance our products and features in markets outside of China, we use our own proprietary model for Doubao, which is available only in China.'
Shortly after The Verge published this article, OpenAI notified The Verge that it had suspended ByteDance's account. OpenAI stated that ByteDance's use of its API was minimal, but that it was suspending the account pending an investigation. If the investigation reveals that ByteDance is not following its policies, OpenAI plans to either force ByteDance to change its usage practices or formally suspend the account.
According to The Verge, while it's not openly discussed, it's common practice for small and medium-sized businesses to use proprietary AI models to help build competing products, especially those from OpenAI. Neither OpenAI nor Microsoft has yet taken any action to crack down on violators, so the use of such APIs is considered a 'legal gray area' among small businesses.
However, those interviewed said it was extremely unusual for a large company like ByteDance to take such measures, which suggests how much pressure the Project Seed development team is under to 'deliver quickly,' The Verge pointed out.
Related Posts:
