TikTok is reportedly developing a clone of its recommendation algorithm tailored for its 170 million U.S. users, potentially allowing for operational independence from its Chinese parent company ByteDance. This initiative, which began late last year, predates legislative efforts in the U.S. to force a sale of TikTok's U.S. operations. The sources, requesting anonymity, suggest that splitting the source code could lay the groundwork for divesting U.S. assets, although there are currently no plans for such action.

The development follows a lawsuit by TikTok and ByteDance challenging the law mandating the sale or ban of the app by January 19. The court set an expedited schedule for considering the legal challenges. Engineers in the U.S. and China have been tasked with separating millions of lines of code, aiming to create an independent code base for TikTok's U.S. operations while removing any links to Chinese users.

The technical separation aims to address bipartisan concerns over potential data access by Beijing. The recommendation engine, crucial to TikTok's popularity, was initially developed by ByteDance engineers in China. The project is described as complex and time-consuming, with estimates suggesting it could take over a year to complete.

Previous efforts, such as Project Texas, failed to satisfy regulators and lawmakers. TikTok executives have discussed options including open-sourcing parts of the algorithm to demonstrate transparency. Compliance and legal considerations complicate the process, as each line of code must be reviewed for inclusion in the separate code base.

The ultimate goal is to create a new source code repository for a recommendation algorithm serving only TikTok U.S., independent of its Chinese counterpart Douyin. However, there's a risk that TikTok U.S. may not match the performance of the existing TikTok due to reliance on ByteDance's engineering resources in China for updates and maintenance.