TikTok and the CCP Propaganda Algorithm

Introduction

TikTok is a short-form video platform owned by ByteDance, a company headquartered in Beijing, China. With over 170 million US users as of early 2025, it is one of the most widely used social media platforms in the United States. Since at least 2019, the app has been subject to national security scrutiny by the Committee on Foreign Investment in the United States (CFIUS), the US Congress, and allied foreign governments.

The concerns are layered and need to be assessed separately: data collection on US users (documented and partially confirmed); algorithm manipulation to suppress or amplify politically sensitive content in the CCP's interest (some evidence, contested interpretation); and a coordinated CCP influence operation using TikTok as a delivery mechanism for propaganda targeting US public opinion (the strongest conspiratorial form — contested and under investigation).

The CFIUS Process and Congressional Action

ByteDance acquired Musical.ly, TikTok's US predecessor, in 2017. CFIUS opened a formal review in 2019, citing national security concerns about data access and potential CCP influence over the platform's operations. CFIUS reviews are confidential; the findings are not published. The Trump administration sought to force a sale in 2020; the Biden administration revived the process; Congress passed the Protecting Americans from Foreign Adversary Controlled Applications Act in April 2024, which required ByteDance to divest TikTok's US operations or face a ban.

TikTok went briefly dark for US users on 18-19 January 2025. President Trump issued an executive order directing the Justice Department not to enforce the ban for 75 days, effectively pausing it. The company and its US operations continued through mid-2025 without a completed divestiture.

The Shou Chew Congressional Testimony

TikTok CEO Shou Zi Chew testified before the House Energy and Commerce Committee on 23 March 2023, and before the Senate in 2024. The testimony revealed several documented facts:

• A "Project Texas" initiative was established to route US user data through Oracle servers to prevent ByteDance employee access.
• Despite Project Texas, ByteDance employees in China had accessed US user data as recently as 2022 — a fact first reported by BuzzFeed News based on leaked internal meetings.
• TikTok's content moderation policies on some political topics differ from US competitors.

Chew denied that ByteDance shares data with the Chinese government and denied that the platform's algorithm is directed by the CCP. Neither claim has been independently verified.

Algorithm Differences: What the Research Shows

Several studies have examined whether TikTok's recommendation algorithm treats politically sensitive topics differently from comparable Western platforms:

• A 2023 analysis found that searches for "Tiananmen Square" on TikTok returned significantly different content than searches on Instagram Reels or YouTube Shorts.
• A separate study by Australian Strategic Policy Institute researchers documented that content related to Xinjiang detention camps was less likely to be amplified by TikTok's algorithm compared to competitor platforms.
• Research by Rutgers University's Network Contagion Research Institute found that TikTok underrepresented content about topics sensitive to the CCP (Tibet, Taiwan independence, Hong Kong protests) compared to Instagram.

These findings are consistent across multiple independent research teams. They show algorithmic differences. They do not, by themselves, prove that the differences are the result of direct CCP instruction rather than ByteDance's own content-policy choices made for business or legal compliance reasons in China.

The Counter-Evidence

Research into whether TikTok actively amplifies pro-CCP propaganda targeting US politics is less conclusive:

• A 2022 RAND study found no consistent evidence that TikTok was being used as a coordinated CCP influence operation comparable to the Internet Research Agency's 2016 operations.
• The Stanford Internet Observatory's analysis of political content on TikTok found polarising US domestic content similar to other platforms, without a clear CCP propaganda signature.
• ByteDance has employed a US-based trust and safety team and has produced transparency reports.

The Genuine Uncertainty

The institutional position — reflected in the CFIUS review, the bipartisan congressional vote, and the executive action — is that the risk is real enough to warrant either divestiture or heightened oversight. This represents institutional concern, not proof of active propaganda operations. The question of whether ByteDance has ever actually served as a propaganda conduit under CCP instruction (rather than simply structuring its platform in CCP-compliant ways for business reasons) is the central contested question, and it has not been resolved as of mid-2025.

Why "Ongoing Investigation"

The data-collection concerns are documented and partially admitted. The algorithmic differences for CCP-sensitive content are supported by multiple independent research studies. The "coordinated CCP propaganda delivery" claim — the strongest form — remains under institutional investigation without conclusive public evidence for or against. This is genuinely unresolved.

Verdict

Ongoing investigation. Documented data-collection concerns and documented algorithmic differences on CCP-sensitive topics are real. Whether these differences represent active CCP propaganda operations rather than ByteDance's own content-policy choices is contested and under investigation. The institutional response (bipartisan legislation, CFIUS review) reflects genuine concern, not resolved proof.

What the Supreme Court Actually Decided in TikTok v. Garland

On January 17, 2025, the US Supreme Court issued a unanimous, unsigned per curiam opinion in TikTok Inc. v. Garland, upholding the Protecting Americans from Foreign Adversary Controlled Applications Act against a First Amendment challenge. It is worth being precise about what the Court did and did not decide. The justices did not rule on whether TikTok's algorithm is, in fact, directed by the Chinese Communist Party, and made no factual finding that ByteDance had transmitted content or propaganda directives to the platform. The ruling addressed a narrower legal question: whether Congress's stated national-security rationale — concern about a foreign adversary's potential access to US user data and potential ability to manipulate content on a platform used by roughly 170 million Americans — was sufficient to justify the law's divest-or-ban structure without violating the First Amendment. The Court found that it was. The law's constitutionality rests on the government's asserted risk, not on a judicial finding that the risk had materialized into an actual propaganda operation. This distinction is frequently collapsed in secondary discussion of the case, which sometimes treats the ruling as though it "proved" CCP control of the algorithm; it did not.

Two NCRI Studies: What They Found and Where They Fall Short of Proof

Beyond the studies already discussed above, Rutgers University's Network Contagion Research Institute (NCRI) has published two specific reports measuring hashtag and search-result disparities between TikTok and rival platforms. The first, "A Tik-Tok-ing Timebomb" (December 21, 2023), compared hashtag volumes between TikTok and Instagram and reported striking gaps: #HongKongProtests appeared roughly 206 times more often on Instagram than on TikTok, and Ukraine-related hashtags were overrepresented on Instagram by roughly 8.5-to-1, well above a 2.2-to-1 baseline the researchers measured for apolitical pop-culture hashtags. The second, "The CCP's Digital Charm Offensive" (August 2024), used 24 accounts built to mimic 16-year-old American users and searched for terms including "Uyghur," "Xinjiang," "Tibet," and "Tiananmen." It found that 60.3% of TikTok's search results for these terms were flagged by the researchers as irrelevant to the query, versus under 5% on comparison platforms — which the authors read as a sign the search algorithm was steering users away from the topics rather than simply underperforming.

Both reports are worth reading alongside their own stated limits. Neither was peer-reviewed. The authors of the 2024 report explicitly characterized their findings as "compelling and strong circumstantial evidence," not proof that the Chinese government or TikTok employees intentionally manipulated the algorithm. The 2023 report's authors removed some of their own COVID-related hashtag comparisons before publication, after acknowledging that TikTok had a documented policy of restricting certain hashtags for anti-Asian-harassment reasons — a reminder that not every hashtag disparity has a CCP-shaped explanation. TikTok, for its part, rejected both reports outright, calling the 2024 study a "non-peer-reviewed, flawed experiment...clearly engineered to reach a false, predetermined conclusion," arguing that hashtags are user-generated rather than platform-assigned, and noting that all recommendation algorithms amplify whatever content is already popular. TikTok separately disabled the hashtag-search tool in its Creative Center dashboard after the reports were published, which researchers said prevented independent replication of the analysis — a fact that undercuts TikTok's transparency claims regardless of which side is right about the underlying numbers.

What Is Actually Confirmed About Who Owns ByteDance

A related but distinct claim — that the Chinese government directly owns or controls ByteDance — is frequently conflated with the algorithm question, and it is worth separating out because it is more directly checkable. According to ByteDance's own disclosed structure, roughly 60% of the company is held by global institutional investors (including major US asset managers), about 20% by its Chinese co-founders, and about 20% by employees, some 7,000 of whom are American. The Chinese government does hold a "golden share" — a 1% equity stake that comes with a board seat and disproportionate voting rights — but that stake sits in Beijing Douyin Information Service Co., a China-only subsidiary that runs Douyin, the domestic sister app to TikTok that operates under Chinese content law. It is not a stake in ByteDance's global parent company, and it has not been shown to extend to TikTok's international operations. This is a case where a real, documented fact — Beijing holds a golden share and a board seat somewhere inside ByteDance's corporate structure — gets frequently overstated into a different and unproven claim: that Beijing owns or directs TikTok globally. Fact-checking organizations and mainstream reporting have made this distinction explicit.

The 2026 Divestiture: Resolved, or Restructured?

The most significant development since the Supreme Court's ruling is that the divestiture actually happened. On January 22, 2026, ByteDance closed a deal establishing the TikTok USDS Joint Venture LLC, transferring roughly 80% ownership to a group of US-based investors — Oracle, Silver Lake, and MGX each taking a 15% stake, with the remainder split among other investors including Michael Dell's family office — while ByteDance retained a minority stake reported at 19.9%. Under the deal's terms, the recommendation algorithm is licensed from ByteDance to the new joint venture rather than transferred outright, and is being retrained using only US user data on Oracle's US cloud infrastructure, with Oracle serving as an independent "trusted technology partner" responsible for auditing security and content-moderation compliance. Adam Presser was named CEO of the new joint venture.

Analysts disagree sharply about whether this resolves the underlying propaganda concern or simply relocates it. The Information Technology and Innovation Foundation argued the deal's structural safeguards — data localization, source-code auditing, independent board oversight, third-party verification — amount to a workable blueprint that eliminates direct CCP control over data and algorithms without requiring an outright ban. The Atlantic Council took a more skeptical view, arguing that because ByteDance still owns the underlying algorithm and merely licenses it, the Chinese government "could exert some influence over how the system evolves," and that the deal redistributes rather than eliminates the risk. Some members of Congress who voted for the original divestiture law raised the same concern: that a licensing arrangement, rather than a clean sale of the algorithm itself, could preserve exactly the dependency the 2024 law was meant to sever. As of this writing, no independent, technical audit results from Oracle's oversight role have been made public, so this remains an assessment of structural risk rather than a demonstrated outcome either way.

Where This Leaves the Claim

Stacking this newer material onto the existing record does not resolve the theory's central, contested claim — it sharpens the boundary between what is documented and what is not. It is documented that TikTok's algorithm produced measurably different results on CCP-sensitive search terms and hashtags across at least two independent, non-peer-reviewed studies, both of which their own authors describe as circumstantial rather than conclusive, and both of which TikTok disputes on methodological grounds while also having limited outside researchers' ability to verify the underlying numbers. It is documented that Congress, the Supreme Court, and the executive branch all treated the underlying risk as credible enough to force a change of ownership, and that the resulting 2026 deal preserves a licensing relationship to a ByteDance-owned algorithm rather than severing it outright — a structural fact that credentialed analysts read in opposite directions. It is not documented, on the public record available as of September 2026, that the CCP has ever issued content directives to TikTok's algorithm team, or that the observed search and hashtag disparities are the result of state instruction rather than ByteDance's own compliance choices, engagement-optimization dynamics, or platform-format differences that could also plausibly explain part of the gap. That gap between a documented pattern and a demonstrated cause is exactly why this theory remains an open investigation rather than a settled finding in either direction.