The Wire
TechnologyPolitics & PolicyLaw & RegulationMedia & Culture

New York says TikTok tested a nonfunctional Algo Refresh feature

New York says TikTok tested a nonfunctional Algo Refresh feature
Photo: techcrunch.com

New York alleges TikTok tested nonfunctional versions of Algo Refresh on young users.

Why it matters: The allegations could increase legal and compliance risks for TikTok's youth-safety controls and recommendation systems. They also raise questions about whether users understood when a safety-related feature was active.

  • New York alleges TikTok ran repeated "ghost" or placebo experiments involving users in 2023.
  • The complaint, as described in the supplied materials, gives no exact number of users; the alleged participants included children and teenagers.
  • New York says some users believed they had activated Algo Refresh even though their recommendations remained unchanged.
  • Reuters reported that the amended complaint became public on October 7, 2026; the supplied materials do not include a separate court-order or docket URL establishing the unsealing.

New York alleges that TikTok conducted repeated "ghost" or placebo experiments in 2023 involving versions of its Algo Refresh feature that did not function as expected. The complaint, as described in the supplied materials, does not provide an exact user count. The alleged participants included children and teenagers. Reuters reported the allegations and the October 7, 2026 unsealing. The supplied materials do not include a separate court order or docket entry for that date.

TikTok describes Algo Refresh as "The option to start fresh on TikTok." The company says enabling it lets users view their For You feed as if they had just signed up, with recommendations no longer based on prior interactions. New York's amended complaint describes the feature.

New York alleges that some users thought they had activated Algo Refresh but received a nonfunctional version while their recommendations remained unchanged. The materials do not establish that those users knowingly enrolled in a placebo experiment or were told a nonfunctional version was active. The state's allegation appears to rest on comparisons between users it says received the functioning feature and users whose recommendations did not change.

New York also alleges that TikTok examined the effects on time spent on the platform and advertising revenue. The materials provide no exact revenue figure and do not specify an exact number of experimental users.

TikTok told Reuters that New York's allegations mischaracterize how Algo Refresh works and said the company "regularly tests new products and features" to understand their real-world performance. The allegations have not been adjudicated.

Attorney General Letitia James sued TikTok on October 8, 2024, alleging that its design harmed young users and that the company misrepresented its safety measures. The case is pending in New York County Supreme Court. New York's 2026 SAFE for Kids Act rules provide broader regulatory context by restricting certain addictive features for minors and limiting overnight notifications for users under 18 without parental consent, but the supplied materials do not tie those rules directly to the alleged 2023 experiments. The attorney general's office announced the original case. The state released the final rules in 2026.

By the numbers

  • 2023 - year New York alleges the experiments occurred
  • October 7, 2026 - date Reuters reported the amended complaint was unsealed
  • October 8, 2024 - date Letitia James announced the original lawsuit

Yes, but: The claims come from New York's amended complaint and have not been adjudicated. The supplied materials do not establish an exact user count, whether participants knowingly joined placebo experiments, or whether they were told a nonfunctional version was active.

What's next: The case remains pending in New York County Supreme Court.

Based on reporting from

  • TechCrunch

See how this story touches your network - open The Wire in Jane.

Open in Jane