TikTok’s Chilling Experiment Exposed

A leaked TikTok report says the company kept a “safer” algorithm away from about 15 million Americans while it watched how much time they would spend inside a more dangerous feed.

Story Snapshot

A confidential TikTok review says 10 percent of United States users stayed on an older, less protected algorithm while a safer version went to everyone else.
The control group reportedly included 16-year-old Chase Nasca, whose feed filled with thousands of suicide and self-harm themed videos before he died by suicide.
The internal document says safety tools “did not take effect…by design,” and that TikTok tracked how the safer feed affected “stickiness” and other engagement measures.
TikTok calls the move a “regular comparative test” and says it has since improved its testing methods, while lawmakers demand answers on how platforms put growth ahead of child safety.

What the leaked TikTok report says happened

In 2021, TikTok engineers built a safer version of the app’s recommendation system meant to keep users from getting trapped in endless loops of harmful content, according to a confidential internal document described by Bloomberg. Instead of giving this system to everyone, the company reportedly set up a giant experiment. About 10 percent of United States users, roughly 15 million people at the time, stayed on the older, less protected version as a control group. The safer feed went to the other 90 percent so TikTok could compare how each group behaved and how much time they spent in the app.

The leaked review says the test was not just about checking if the new safety rules worked. It also tracked whether breaking up “filter bubbles” would reduce the app’s “stickiness,” a term companies use for how hard it is for people to put the product down. That choice strikes a nerve for many Americans who already feel that big tech firms treat them as test subjects, not customers, and value extra minutes of screen time more than the mental health of their kids. The report’s language that protections were withheld “by design” has become a flashpoint in the debate.

The tragic teen case at the center of the story

The internal report reportedly shows that this control group included 16-year-old Long Island student Chase Nasca, who died by suicide in February 2022. After his death, TikTok staff reconstructed his feed and analyzed more than 7,500 videos the system had shown him in his last two weeks of life. According to those reviews, a large majority of the clips centered on depression, loneliness, suicide, and other mental health struggles. The document says his account fell into a “filter bubble,” where the algorithm kept serving similar dark content it believed he would keep watching.

Because Chase’s account was part of the control group, the company’s own review says its “filter bubble prevention strategies did not take effect on this user by design.” His parents later sued TikTok, arguing the app’s feed flooded their son with self-harm and hopelessness even though he did not search for that material. The available reporting does not prove that the algorithm caused his death, and no court has made that finding. But the case shows in painful detail how a recommendation system, once tuned toward engagement, can trap a teen in a spiral of despairing content in the weeks before a tragedy.

How TikTok defends the test — and why both parties are alarmed

TikTok has pushed back on the harshest claims, saying through a statement reported by international media that the experiment was a “regular comparative test to verify safety features” and that it has since tightened its testing methods. Another outlet notes that Bloomberg’s story does not prove the company meant to expose users to harm, only that it kept some people on the old system to measure changes in engagement. Even so, the company has not publicly released the full internal report, the original test plan, or the detailed safety review that would show exactly who approved the design and what risks they flagged.

TikTok withheld an algorithm safety update from millions of people in the United States while it studied how the change would affect user engagement – reporthttps://t.co/SqwiMlFQZz

— Insider Paper (@TheInsiderPaper) August 6, 2026

Lawmakers from both parties see this as another sign that powerful platforms and government watchdogs have both failed families. A letter from Senator Edward Markey and colleagues to TikTok’s United States Data Security unit, sent in the wake of a separate 2025 deal, already warns that the company’s recommendation engine must be closely watched for abuse and harm to Americans. Many conservatives worry that a foreign-owned app is shaping the minds of United States teens while chasing profit. Many liberals fear that tech giants are allowed to run human experiments on children with almost no real oversight. Both sides share a deeper anger that, once again, unelected corporate and bureaucratic elites made secret choices about kids’ safety and only admitted them after a leak.