New Mexico judge orders Meta to remove Like counts and cap teen use at 90 hours a month

Courts stopped talking about content and started talking about design. That is the shift you need to track. A New Mexico judge did not just fine Meta for bad posts. He named specific product features, Like counts, push notifications, session limits, and ordered them changed. Meanwhile, TikTok's own A/B test of a safety feature is now legal evidence. Let me catch you up on what that means for the decisions your team makes every week.

The deep cut

• Design choices are now legal exposure. A New Mexico judge ordered Meta to remove Like counts and cap teen usage at 90 hours a month.
• Safety tests need a real ceiling on the control group. TikTok left 15 million users off a safety fix to protect its engagement metrics.
• Your public safety claims become the standard you are judged against. Shou Chew told Congress teen safety was "a core priority" while TikTok's own review sat unmentioned.

When the ruling reaches into your feature list

The New Mexico penalty is big, nearly $942 million total, but the number is not the news for you. The order tells Meta how to build. It must remove Like counts for New Mexico teens, show those metrics only with parent approval, pause push notifications to minors between 10 p.m. and 7 a.m., and cap teen usage at 90 hours a month.

That is a court writing a product spec. The judge called Meta a "public nuisance" and treated engagement features as the cause of harm, not the content flowing through them.

Attorney General Raul Torrez put the theory plainly: Meta "chose engagement and profit over their safety." That is the frame courts are testing now. The design decision itself is on trial.

The A/B test that became evidence

Every product team runs experiments. TikTok ran one that should scare yours. To learn if breaking up harmful filter bubbles would hurt engagement, TikTok held 10 percent of U.S. users on the old system, roughly 15 million people, with no safety fix. The internal report called it a "delicate balance" between user safety and measuring the metric.

One person in that control group was Chase Nasca, a 16-year-old who died by suicide. TikTok later found 73 percent of the 7,563 videos served to him in his final two weeks carried themes of suicide, sadness, or self-harm.

The lesson is not "stop testing." It is that a safety holdback is different from a checkout-button holdback. When TikTok proposed shrinking the control group, it still left 1.8 million users exposed. Put a hard ceiling and a short clock on any test where the withheld thing protects people.

What you say in public is the bar you get held to

TikTok told Congress it had "a commitment to transparency," and CEO Shou Chew called teen safety "a core priority." Two weeks before that hearing, the company finished the internal review of Nasca's feed. When lawmakers asked TikTok to list every case where A/B testing harmed users, it named none and did not mention the experiment at all.

Meta is running the same play now, saying it remains "confident in our record of protecting teens online." Courts read those statements against what your logs show.

The gap between your public safety promise and your internal metrics is the thing that turns into a headline and a fine. Write your press lines like a plaintiff will read them next to your dashboards.

The team you cut is the team you point to later

TikTok is laying off 250 people and closing its Nashville office, which held part of its content moderation staff, and leaning harder on AI to catch harmful content. The timing is rough. Days earlier, TikTok blamed a "moderator error" for the slow takedown of a Perez Hilton livestream showing self-harm.

Thinning the safety org while courts assign dollar penalties to safety failures is a bet against the trend. When something goes wrong, the defense rests on the people and systems you funded. Cut them quietly and the record shows it.

Three questions for your team

• Which of our current features, Like counts, infinite scroll, streaks, late-night notifications, would a judge be able to name and order us to change? Write that list before someone else does.
• On our next safety-related experiment, what is the hard cap on the control group size and how many days does it run? If we cannot answer both, we do not ship the test.
• Do our public safety claims match what our own metrics and internal reviews would show a court? Have someone outside the comms team check.