New York Says TikTok Gave Teens a Placebo Safety Reset
New York alleges TikTok ran an experiment that gave thousands of teenagers and children a nonworking version of a safety feature they had requested. The claim was detailed in a state lawsuit and reported on Oct. 8, 2026 TechCrunch.
The feature is Algo Refresh. It lets users reset their recommendations and stop seeing unwanted or potentially harmful content. For users trying to leave a loop of distressing videos, it is presented as a hard reset that clears personalization state, the stored history that shapes the feed, and starts over.
According to the allegation, users in the test group believed they had activated Algo Refresh but their feeds did not change. They had opted in. A separate group received a working version, and only that group saw recommendations reset as described. In testing terms, New York is describing a placebo control applied not to ranking tweaks or interface wording, but to a safety control the user chose to use.
TikTok said it "regularly tests new products and features to validate the experience and understand how they work in the real world." That approach matches common product testing, where holdout groups and different variants are used to measure cause and effect before a full rollout.
The lawsuit also points to a separate TikTok experiment that withheld a safety feature from 15 million U.S. users, including a teenager who later died by suicide. The verified facts do not establish a causal link between the withheld feature and that death. They establish the size of the test and that the teenager was included in it.
The New York case is part of a larger state-level legal campaign. It is one of more than two dozen state cases accusing TikTok of designing its platform to encourage addictive use among children and misleading users about safety. That campaign includes a coordinated action filed in October 2024 by 13 states and the District of Columbia over accusations of harming younger users Reuters. Alabama has reached a settlement with TikTok and ByteDance over teen safety claims. ByteDance is TikTok's Chinese parent company.
The broader context here will be familiar to people who follow recommender systems. A/B testing, staged rollouts, and holdouts are basic infrastructure. Teams use them to check whether a change to ranking, retrieval, or user controls actually changes viewing behavior, retention, or reported harm. The allegation in New York concerns where that infrastructure was used and what users were told when they pressed the button.
In my view, that placement matters more than the sample size. Testing an unproven ranking change on a small slice of traffic with internal monitoring is routine engineering. Testing a labeled safety reset while leaving personalization in place, with the screen still signaling success, is a different case. The technical question is auditability. A reset implies clear effects on stored user profiles, watch-history signals, or filtering of candidates. If those effects can be quietly turned off for one test group, outside researchers and regulators cannot verify what the control does without measuring both the app and the feed.
From a builder's perspective, there is a disclosure problem that deserves attention. Experiment systems normally log who saw which version and track safety metrics internally. They rarely show that assignment to the user. When the feature is about safety, that silence carries legal and ethical weight. Users seeking relief from unwanted content are acting on the belief the system has changed. An internal log of no change does not correct the message on screen.
Looking over the longer term, clearer design for safety controls would help both users and builders. Versioned reset rules that can be inspected, plain confirmation of what was cleared, and paths for independent checks would make safety claims testable rather than asserted. That would not end lawsuits over youth safety, but it would narrow disputes to behavior that can be measured rather than competing accounts of what an experiment did.
