New York alleges TikTok gave teens, children a placebo safety feature instead of a real one

What is TikTok’s “Algo Refresh” feature, and what is it supposed to do?

“Algo Refresh” is described as a safety feature that lets users reset their recommendations. Its purpose is to change the content selected for their feeds and stop unwanted or potentially harmful material from continuing to appear. That matters because users may rely on the feature when they feel their feed is unsafe or unsuitable.

In the experiments described in New York’s lawsuit, some users believed they had activated Algo Refresh. However, their feeds did not change. Users outside the experiment had access to a fully functioning version, creating a difference between what some participants expected and what they actually received.

The allegations are part of a wider dispute over TikTok’s treatment of young users. New York and more than two dozen other states accuse TikTok of misleading users about safety and encouraging addictive use. TikTok says it regularly tests products and features to understand how they work in the real world.

How many people were affected by the experiments, including the reported 15 million U.S. users in one test?

Reuters reported that TikTok’s experiments involved thousands of users, including teens and children. The article separately describes a Bloomberg report about another experiment that withheld a safety feature from 15 million U.S. users. These figures show the reported tests could affect both selected groups and a very large portion of TikTok’s U.S. audience.

The article does not say that all 15 million people were part of the same experiment as the thousands who used Algo Refresh. It presents the 15 million figure as belonging to another test. For that reason, the available figures cannot be combined into one total without risking an inaccurate count.

The scale is central to the legal dispute. New York’s lawsuit is one of more than two dozen state cases accusing TikTok of misleading users about safety and designing its platform to encourage addictive use among children. TikTok says it regularly tests features to understand real-world results.

What was different between users who received the real safety feature and those who received the placebo version?

The key difference was whether the safety feature actually worked. Users who were not part of the experiment had access to a fully functioning version of Algo Refresh. Users included in the experiment believed they had activated the tool, but the recommendations shown in their feeds did not change.

That made the experiment’s version a placebo, according to the lawsuit’s allegations. A placebo version can look like a real feature to the user while leaving the underlying result unchanged. Here, the expected result was a refreshed feed with unwanted or potentially harmful content no longer appearing in the same way.

This difference matters because users may make safety decisions based on what a platform tells them a tool has done. The lawsuit claims TikTok conducted experiments involving thousands of users, including children and teens. New York and other states say the company misled users about safety, while TikTok says it tests products and features to assess real-world performance.

What consequences could users face if they believed unwanted or harmful content had been removed when their recommendations had not actually changed?

If users believed Algo Refresh had worked, they might assume their recommendations were safer or no longer reflected unwanted content. But the allegations say some feeds did not change. The immediate risk was a gap between the protection users expected and the protection they actually received.

For example, a teenager could activate the feature after seeing content they wanted to avoid, then continue receiving similar recommendations without realizing the reset had not occurred. The article does not establish a specific injury caused by this mismatch. It does report that one other experiment withheld a safety feature from 15 million U.S. users, including a teenager who later died by suicide.

The broader consequence is reduced trust in safety controls. New York and other states accuse TikTok of misleading users and encouraging addictive use among children. The allegations could increase pressure for clearer testing, reliable safety tools, and greater accountability, although the article does not state what legal remedies will follow.

Why are New York and other states suing TikTok over its treatment of children and teenagers?

New York is suing TikTok because the state alleges that the platform was designed to encourage addictive use among children. The lawsuit also accuses the company of misleading users about safety features. These claims place both TikTok’s product design and its communication with users under scrutiny.

The allegations include experiments involving thousands of users, including teens and children. In one reported test, participants believed they had activated Algo Refresh, but their recommendations did not change. Bloomberg also reported that another experiment withheld a safety feature from 15 million U.S. users, including a teenager who later died by suicide. The article does not establish that the withheld feature caused that death.

New York’s case is one of more than two dozen lawsuits brought by states. Together, they reflect a broader challenge to TikTok’s treatment of young users. TikTok says it regularly tests products and features to validate experiences and understand how they work in the real world.

What protections do users normally expect when a technology company tests a product feature on them?

When a technology company tests a feature on users, people generally expect honest disclosure about the test and a clear understanding of what the feature will do. They also expect the feature to provide the protection promised, especially when it is presented as a safety tool. These expectations matter because users may change their behavior based on the feature’s apparent operation.

In this case, the allegations say users believed they had activated Algo Refresh, but their feeds did not change. Users outside the experiment received a functioning version. That contrast suggests participants may not have known that the tool they saw was not producing the expected result, though the article does not describe the experiment’s consent process or notices.

The article gives TikTok’s explanation that it regularly tests products and features to validate experiences and understand real-world performance. It does not identify specific safeguards, such as opt-outs, warnings, or independent oversight. Those details would be needed to judge whether normal protections were provided.

How do recommendation algorithms learn what content to show, and why can repeated viewing contribute to extended or addictive use?

Recommendation algorithms generally learn from user activity, such as the content someone watches, skips, likes, shares, or follows. They compare those signals with patterns from other users and predict which posts may hold a person’s attention. The source article does not explain TikTok’s algorithm, so this description uses established knowledge about recommendation systems.

Repeated viewing can strengthen the system’s belief that similar material interests the user. If each recommendation leads to another view or interaction, the feedback loop can keep producing related content. A user may continue scrolling because each new item is selected to match earlier behavior. This can contribute to longer or more repetitive sessions, although the article does not detail the algorithm’s technical design.

That mechanism helps explain why the lawsuits focus on platform design and addictive use among children. If recommendations repeatedly encourage continued viewing, young users may find it harder to stop. The article connects the legal claims to TikTok’s design and safety practices, while TikTok says it tests features to understand how they work in real-world use.

Key Facts:

📌 Algo Refresh lets users reset their TikTok recommendations.

📌 The tool is meant to stop unwanted or potentially harmful content.

📌 Some users believed they activated it, but their feeds did not change.

📌 Reuters reported experiments involving thousands of users.

📌 One other test allegedly withheld a safety feature from 15 million U.S. users.

📌 The article describes these as separate experiments.

📌 Nonparticipants had access to a fully functioning Algo Refresh tool.