OpenAI has suspended the account of TikTok parent ByteDance following a report that the China-owned firm was “secretly” using its “GPT” AI technology to build its own rival chatbot service.
ByteDance, whose ties to the Chinese government have prompted intense federal scrutiny of TikTok’s operations in the US, has reportedly relied on OpenAI’s application programming interface, or API, “during nearly every phase of development” of its AI product, The Verge reported, citing internal company documents.
ByteDance has been building its own large-language model – an AI tool trained on endless reams of internet data – as part of a “high-priority” and “secretive” internal initiative called “Project Seed.” The report said ByteDance has used OpenAI’s “GPT” to both train its model and evaluate its development.
The practice appears to violate the stated terms of service for OpenAI, which tells customers its AI platform can’t be used “to develop any artificial intelligence models that compete with our products and services.”
“All API customers must adhere to our usage policies to ensure that our technology is used for good,” an OpenAI spokesperson told the outlet. “While ByteDance’s use of our API was minimal, we have suspended their account while we further investigate.”
“If we discover that their usage doesn’t follow these policies, we will ask them to make necessary changes or terminate their account,” the spokesperson added.
ByteDance reportedly buys its access to OpenAI’s tools through Microsoft, the AI firm’s key investor and partner. Microsoft has similar terms of service.
The Verge said it has seen internal discussions on ByteDance’s “Lark” messaging service in which employees chatted about how to “whitewash” evidence of the company’s use of OpenAI. ByteDance are said to have regularly hit usage limits for the service.
ByteDance reportedly received regulatory approval in China to launch its AI chatbot, dubbed “Duobao,” last August. The company reportedly told its team to stop using OpenAI-generated text within that same timeframe.
“They say they want to make sure everything is legal, but they really just don’t want to get caught,” a source with knowledge of ByteDance’s inner workings told The Verge.
Duobao is already available in China and it’s operated via China-based servers. It is not currently available on TikTok.
When reached for comment, ByteDance representative Jodi Seth reportedly confirmed that company engineers had used OpenAI tools during the early stages of Project Seed – but asserted any related text was yanked out of the chatbot’s data set earlier this year.
“ByteDance is licensed by Microsoft to use the GPT APIs,” Seth said in a statement. “We use GPT to power products and features in non-China markets, but use our self-developed model to power Doubao, which is available only in China.”
Meanwhile, a Microsoft spokesperson told the Verge that the company requires all AI customers to be approved for access.
“We also set standards and provide resources to help our customers use these technologies responsibly and in compliance with our terms of service, and have processes in place to detect misuse and discontinue access if companies violate our code of conduct,” the spokesperson said.
The Post has reached out to OpenAI, Microsoft and ByteDance for additional comment.
ByteDance’s apparent bid to become an AI leader could reignite concerns about the company’s operations.
Last May, a former US-based ByteDance executive alleged that the Chinese Communist Party “maintained supreme access” to all data held by TikTok parent, including that which was stored in the US. ByteDance slammed the allegations as “baseless” and denied wrongdoing.
In March, the Biden administration reportedly warned ByteDance executives that TikTok would face a US ban unless the company sold its stake. Global investors own about 60% of ByteDance shares, according to the company, while 20% are held by employees and 20% by its founders.
Meanwhile, Republican lawmakers have stepped up calls for a TikTok ban due to concerns about the spread of antisemitism and anti-Israel content on the platform. The company has denied wrongdoing and said it has aggressively cracked down on posts that violate its rules.
This story originally appeared on NYPost
