After a U.S.-led investor group headed by Oracle’s Larry Ellison took control of TikTok’s American business, users accused the app of throttling videos about ICE raids, the late sex offender Jeffrey Epstein, the fatal shooting of Alex Pretti in Minneapolis, and other politically charged topics. The hashtag #TikTokCensorship trended on X, some users switched to alternative apps, and state and international officials called for investigations.
A new analysis published by Good Authority and conducted by eight academics offers a different account. Using view metrics from more than 100,000 videos, the team tracked how often TikTok recommended posts about ICE, Pretti, Renee Good (a woman killed by an ICE agent), and the keywords Trump and Epstein. They compared those trends to nonpolitical subjects such as food recipes and Oscars coverage.
The researchers found a sharp decline in views across all topics around the time of a TikTok data-center outage. Total views plunged and then began to recover after service resumed. Those patterns, the study argues, are consistent with a broad technical disruption rather than an organized, top-down political purge.
However, the authors are careful not to rule out other possibilities. Their analysis does not find evidence of systemic political censorship based on the available public data, but small-scale removals, shadowbanning of individual accounts, or targeted moderation could escape detection in aggregate trends. Private direct messages, where some users reported blocked words, are not accessible for outside study, so those claims cannot be evaluated in this dataset.
A central limitation is the lack of researcher access to TikTok’s internal systems. The platform does not provide the level of transparency needed to trace how its recommender shapes amplification and suppression, the researchers say. They urge TikTok and other platforms to enable third-party study of recommender systems to look for undue political influence.
Timing amplified anxieties: Ellison is a known ally of former President Trump, and critics worry new ownership might reshape content moderation. Anupam Chander, a law and technology professor at Georgetown, said the new owners will have to earn public trust and could demonstrate neutrality by welcoming independent researchers and diverse hires.
Ownership of the U.S. entity includes Oracle, private equity firm Silver Lake, and Emirati investor MGX. ByteDance, TikTok’s Chinese parent, retains a minority stake and still owns the algorithm, which is to be retrained on American users’ data. Backers of the U.S. company say China will not control the algorithm and that Oracle will supervise it, but skeptics note that the underlying technology remains tied to Beijing.
A TikTok spokeswoman told reporters the platform has made no changes to its algorithm since the new investors took control of the American business. As Benjamin Guinaudeau, a study author and professor at Université Laval, put it: TikTok can assert things about algorithmic changes, but outside observers cannot verify them. Major, obvious shifts in content would likely be detectable; subtle adjustments in the ‘For You’ recommender, he warned, are nearly impossible to spot without more extensive internal data.