New York alleges that TikTok gave thousands of users—including children and teenagers—nonfunctional versions of its “Algo Refresh” safety control during repeated tests in 2023. The claims appear in an amended state lawsuit that was filed under seal in August 2026 and later ordered unsealed. TikTok says the allegations mischaracterize how the feature works and describes product testing as a normal way to validate user experiences.
This is an allegation in active litigation, not a final judicial finding. It also does not establish that TikTok’s current refresh tool is nonfunctional for users today. The newly public claims instead concern how particular test groups were treated and what TikTok allegedly told people about the control.
What New York alleges about Algo Refresh
Reuters reported on October 7 that New York’s complaint describes “ghost” or placebo experiments involving users who tried to activate safety features. In the alleged Algo Refresh tests, some people believed they had reset their recommendations, but their feeds did not change as they did for users who received the working version.
The state alleges TikTok repeated the tests with thousands of users in 2023 and compared the nonfunctional group with people who received the tool. According to the complaint as reported by Reuters, the company studied effects including time spent on TikTok and advertising revenue.
New York also alleges that a TikTok Digital Well Being product manager raised an internal objection because a placebo group conflicted with the feature’s stated purpose of algorithm transparency and user control. The Verge separately reported the unsealed allegations and TikTok’s response on October 8.
What TikTok says
TikTok did not accept New York’s characterization. A company spokesperson told Reuters and The Verge that TikTok regularly tests products and features as it improves the platform, seeking to understand how they work in real-world use. Reuters reported that TikTok has broadly denied New York’s claims in court and says protecting younger users is a priority.
The company also said the complaint mischaracterizes Algo Refresh. That response matters because the litigation has not produced a final ruling on the new allegations, and public reports do not provide the complete evidentiary record that a court may eventually consider.
What the TikTok refresh feature is supposed to do
TikTok publicly introduced the refresh option on March 16, 2023. In its official announcement, the company said activating it should make the For You feed behave as though the user had just joined TikTok. The recommendation system would then begin learning again from new interactions.
TikTok’s current support instructions direct users to Profile, Menu, Settings and privacy, Content preferences, and then “Refresh your For You feed.” The company says refreshing affects only For You recommendations. It does not unfollow accounts or override other content settings, and the action cannot be undone after confirmation.
New York’s case focuses on whether people in the alleged test groups received the control they reasonably believed they were activating—and whether TikTok accurately described what the reset accomplished.
What users can and cannot conclude
Users cannot tell from the lawsuit whether their own current refresh request works. The reports identify tests conducted in 2023, not a present-day systemwide failure. Neither New York nor the reporting says every user received a placebo version.
The allegations do raise a broader question about informed consent in platform experiments. A normal interface experiment might compare layouts or button colors. A test involving a safety or control feature is more consequential when users intentionally seek relief from unwanted recommendations and believe their choice has taken effect.
Anyone trying to reshape a difficult For You feed can use several controls together rather than depending on one action. TikTok offers “Not interested,” keyword filters, topic controls and the Following feed in addition to refresh. Users can also review their watch and search behavior because recommendation systems learn from continued interactions after a reset.
Why creators and brands should pay attention
Recommendation controls affect more than individual viewing. They influence which creators can be discovered, which topics are repeatedly amplified and how audiences understand the platform’s promises. If a user explicitly asks to reset recommendations, creators and advertisers have an interest in knowing whether subsequent reach comes from a genuinely retrained feed.
Brands should avoid treating the lawsuit as proof of current manipulation or using it for opportunistic attacks on competitors. The responsible approach is to monitor the case, preserve accurate campaign records and separate verified platform changes from allegations. That same distinction was important in The Daily Vantage’s report on Meta’s restriction on TikTok-linked ads, where paid advertising and organic posting required separate analysis.
Creators should likewise avoid telling followers that the refresh button is “fake.” The available evidence supports a narrower statement: New York alleges that selected users received nonfunctional versions during 2023 experiments, while TikTok disputes the state’s framing.
How this differs from TikTok commerce updates
This case concerns recommendation control and platform safety, not TikTok Shop transactions or creator monetization. Readers looking for the separate commerce change can review how the TikTok Shopping Assistant and Buy Direct features work. Combining the two issues would obscure both the legal allegation and the practical shopping tools.
What happens next
The amended complaint is part of New York’s broader lawsuit accusing TikTok of misleading the public about youth safety and designing features that encourage compulsive use. TikTok contests the claims. The newly unsealed material may now be examined in litigation, but publication of an allegation does not determine liability.
For now, the verified conclusion is limited: New York says TikTok withheld a working recommendation-reset tool from some users during placebo-style tests; TikTok says routine testing has been mischaracterized. Whether those practices violated the law remains for the court to decide.
.jpg){kind=link}