TikTok Plans to Build US-Focused Recommendation Algorithm Amid Political Pressure

TikTok has revealed plans to construct a duplicate of its recommendation algorithm tailored specifically for its 170 million American users.
 This distinct system may lay the groundwork for an independent version of the app in the US, potentially addressing concerns raised by lawmakers who have advocated for a ban, as per sources familiar with the company's efforts.

The initiative to segregate the algorithm's code base commenced prior to the passing of a US law in April mandating the sale of TikTok's US operations.

 Despite the company's denial of any divestiture plans, anonymous sources have hinted that the new algorithm could potentially facilitate such a move.

Initially, TikTok declined to comment on the matter, but later refuted the report's accuracy without specifying the inaccuracies. 

The company has previously stated that a forced sale would be unfeasible, and a legal challenge to the law is currently undergoing expedited review.

The project involved the meticulous division of millions of lines of code by hundreds of engineers in the US and China to create a system independent from the Chinese version of TikTok, known as Douyin. This separation aims to eliminate any ties to Chinese user data.

This development sheds light on the potential technical complexities associated with separating TikTok's US operations and underscores the company's efforts to navigate political pressures. US officials, including President Biden, have expressed concerns about the security of user data and the potential influence of the Chinese government.

While previous reports indicated that a complete sale of the app, including the algorithm, is improbable due to Chinese export controls, the core code of the recommendation engine originated in China but was adapted for various global markets, including the US. TikTok has attributed much of its success to the effectiveness of this user-focused algorithm.

Post a Comment

Previous Post Next Post