May 30 (Reuters) – TikTok is creating a separate recommendation algorithm for its 170 million U.S. users, aiming to operate independently from its Chinese parent company, ByteDance, to address concerns from American lawmakers who have been pushing for a ban.
Thank you for reading this post, don't forget to subscribe!Sources with direct knowledge of the efforts revealed that ByteDance ordered the separation of the source code late last year, predating a bill that gained momentum in Congress this year, which mandates the sale of TikTok’s U.S. operations. The bill was signed into law in April.
Sources, who requested anonymity due to the sensitivity of the information, indicated that splitting the code could potentially set the stage for divestiture of TikTok’s U.S. assets, though there are currently no plans for such a sale. TikTok has previously stated that selling its U.S. assets would be impossible.
After Reuters published the story, TikTok responded on X, stating, “The Reuters story published today is misleading and factually inaccurate,” without specifying the inaccuracies. A Reuters spokesperson maintained the accuracy of their reporting.
TikTok and ByteDance filed a lawsuit in U.S. federal court in May, challenging the law that requires a sale or ban of the app by January 19. A U.S. appeals court has scheduled an expedited review of the legal challenges to this new law.
Over the past few months, hundreds of engineers from ByteDance and TikTok in the U.S. and China have been tasked with separating millions of lines of code. Their goal is to create an independent code base for the U.S. version of TikTok, separate from the Chinese version, Douyin. This separation involves removing any links to Chinese users.
This complex and meticulous task is expected to take more than a year. The effort underscores TikTok’s attempts to mitigate bipartisan political risks, with U.S. President Biden and other proponents of the law expressing concerns over TikTok’s data practices.
In 2020, the Chinese government added content recommendation algorithms to its export-control list, complicating any potential sale of TikTok’s algorithm. ByteDance attributes TikTok’s success to its effective recommendation engine, which customizes user feeds based on their interactions.
The separation of TikTok’s U.S. operations from ByteDance involves significant compliance and legal challenges. Each line of code must be reviewed to determine if it can be included in the new, independent code base. The objective is to establish a new repository for a recommendation algorithm that serves only TikTok U.S., operating independently from TikTok apps in other regions and Douyin.
Despite these efforts, TikTok management acknowledges the risk that the U.S. version may not match the performance of the existing platform, as it currently relies on the engineering capabilities of ByteDance’s team in China to maximize user engagement.
