TikTok Sale Leaves ByteDance’s Algorithm Role Unclear

TikTok’s ownership battle was supposed to have a clear ending. Congress passed a law forcing ByteDance to sell the platform or lose access to the U.S. market. The Supreme Court unanimously upheld that law. After months of negotiations, American investors took control of TikTok’s U.S. operations, allowing President Donald Trump to declare the dispute resolved.

But ownership was never the only thing Congress said it was worried about.

The debate that consumed Washington wasn’t simply about whether TikTok was Chinese-owned. It was about whether the recommendation system deciding what roughly 170 million Americans watch every day could still be influenced by ByteDance.

Six months after the deal closed, that question remains surprisingly difficult to answer not because nobody has asked it, but because almost none of the technical details needed to answer it have been made public.

Ownership changed, But the algorithm became the harder question.

The transaction itself appears straightforward on paper. TikTok’s U.S. operations now sit inside TikTok USDS Joint Venture LLC, with 80.1% owned by American investors, including Oracle, Silver Lake, MGX, Michael Dell’s family office and several other investment groups. ByteDance retained 19.9%, keeping it below the ownership threshold established under the 2024 divestiture law.

That headline number dominated most coverage of the agreement. The smaller detail buried inside the legislation attracted far less attention.

Congress didn’t simply require ByteDance to reduce its ownership. The law also prohibited “any cooperation with respect to the operation of a content recommendation algorithm” between ByteDance and the new U.S. company. In other words, lawmakers weren’t only trying to change who owned TikTok. They were trying to separate the recommendation engine from the company that built it.

The final agreement leaves that separation difficult to evaluate from the outside. ByteDance continues licensing the underlying recommendation technology while Oracle is responsible for retraining the system using U.S. user data.

Beyond those broad descriptions, neither the companies nor regulators have explained how that process works, who audits it or what technical safeguards prevent influence over future updates. For legislation built around algorithmic control, remarkably little has been published about the algorithm itself.

Researchers say the core concern hasn’t disappeared

That lack of transparency is why many researchers remain cautious about declaring the issue settled.

Christo Wilson, a computer science professor at Northeastern University who studies recommendation systems, has argued that the ownership structure answers only part of the original concern. The recommendation engine itself remains proprietary, making independent verification almost impossible.

“The algorithm is still a proprietary black box,” Wilson said. “No one really knows how the algorithm works. That is as true today as it was when this all started.”

His observation reflects a broader challenge facing governments trying to regulate AI-driven platforms. Ownership structures can be disclosed through corporate filings. Recommendation systems cannot. Unless outside experts are allowed to inspect how ranking decisions are made and updated over time, the public has little way of knowing whether meaningful control has actually changed.

That uncertainty doesn’t prove ByteDance still influences recommendations. It simply means the public cannot independently verify that it doesn’t.

Congress won the legal battle, then largely left the stage

The political response after the deal has been almost as striking as the technical questions.

The original legislation passed with overwhelming bipartisan support, with lawmakers repeatedly arguing that TikTok’s recommendation engine represented the central national security concern. Yet after the joint venture was announced, congressional oversight largely faded from public view.

Sen. Ed Markey has remained one of the few lawmakers continuing to question the arrangement, saying the finalized agreement “raises many more questions than answers” and criticizing the administration for providing “virtually no details” despite repeated requests.

That silence creates an uncomfortable gap.

Either lawmakers received technical assurances that have never been disclosed publicly, or attention shifted once the ownership issue appeared politically resolved. Outside government, there is no practical way to distinguish between those possibilities.

The first test arrived almost immediately

The new ownership structure also faced scrutiny sooner than many expected.

Within days of the transaction closing, California Governor Gavin Newsom launched an investigation into allegations that TikTok had suppressed content critical of President Trump. Whether those claims ultimately prove accurate remains an open question, but their timing highlights something important.

The accusations surrounding TikTok changed, but the underlying concern did not. Before the sale, critics worried about Chinese influence over recommendations. After the sale, critics questioned whether political influence could emerge under a different ownership structure.

In both cases, the debate centered on the same issue: a recommendation system that outsiders cannot independently audit. The ownership dispute may have ended. The transparency debate clearly has not.

The next phase won’t be about ownership

The TikTok fight increasingly looks less like a dispute over corporate control and more like an early test of how governments regulate algorithmic systems.

Requiring companies to change ownership is relatively straightforward. Requiring them to demonstrate how complex recommendation engines operate is significantly harder. Algorithms evolve continuously through retraining, ranking adjustments and new data inputs changes that cannot be understood simply by looking at a shareholder register.

That challenge extends well beyond TikTok. Governments in the United States and Europe are beginning to debate whether independent algorithmic audits should become part of future technology regulation, particularly for platforms capable of shaping public discourse at enormous scale.

The TikTok deal may ultimately be remembered less for ending a political battle than for exposing the limits of ownership as a regulatory solution.

Why it matters

Congress succeeded in forcing ByteDance to relinquish majority ownership of TikTok’s U.S. business. The Supreme Court confirmed lawmakers had the authority to do so. The joint venture delivered the structural outcome Washington demanded.

What none of those milestones settled was the question that drove the legislation in the first place: who ultimately controls the system deciding what millions of Americans see each day.

Until that answer can be verified rather than assumed, TikTok’s ownership battle is better understood as a change in corporate structure than a final resolution to the algorithm debate.