A Federal Court Just Drew a Line: Social Media Algorithms Are Not Untouchable

Publish Date : 08/23/2026

For years, social media companies have defended the systems used to determine what children see, how long they remain online, and what appears next on their screens as an integral part of their platforms.

California is now challenging the model.

In a significant ruling issued on August 21, 2026, a federal judge refused to stop the state from enforcing portions of a law designed specifically to protect children from what California calls “addictive feeds.”

The companies fighting the law are some of the largest technology companies in the world: Meta, TikTok, Google and YouTube.

Their argument, at its core, is remarkable.

They contend that the algorithms used to select and recommend content to users are a form of protected expression under the First Amendment.

U.S. District Judge Edward J. Davila was not persuaded.

The court concluded using a predictive algorithm to determine what content a user is likely to find “interesting” is non-expressive conduct and therefore not protected by the First Amendment in the manner the companies contend.

That distinction could have consequences far beyond one California statute.

Because underneath the constitutional arguments is a much larger question that parents, regulators and courts increasingly are being forced to confront:

What happens when a social media platform is not simply hosting content, but actively deciding, through algorithms designed around a user’s behavior, what a child will see next?

California’s Attempt to Regulate the Addictive Feed

California Senate Bill 976 is formally known as the Protecting Our Kids from Social Media Addiction Act.

Importantly, the law does not prohibit children from using social media. It does not ban TikTok, Instagram or YouTube for minors and it does not prevent parents from deciding whether their children should be permitted to use these platforms.

Instead, California targets features of the platforms, including the personalized recommendation systems continuously selecting content for users.

The statute defines an “addictive feed” to include an online service or portion of a service where multiple pieces of user-generated or user-shared media are selected for display based upon information associated with the user or the user’s device.

In simpler terms, the law targets technology behind the seemingly endless stream of personalized content.

Watch one video – The platform learns.

Pause on another – The platform learns.

Search for something – The platform learns

Like something – The platform learns

Skip something – The platform learns.

Return to something – The platform learns.

The system keeps learning.

The next piece of content is then selected using information the platform has accumulated about the user.

For an adult, that technology raises significant questions.

For a child, California concluded it raises something more serious: a potential threat to children’s mental and physical health.

What S.B. 976 Requires

Among other things, the challenged provisions restrict platforms from providing these personalized “addictive feeds” to minors without verifiable parental consent.

The law also requires platforms to develop settings giving parents greater control over their children’s social media use. This is an important distinction.

California did not tell social media companies what opinions people may express.

It did not prohibit creators from uploading videos.

It did not eliminate direct messaging.

It targeted the machinery deciding what content gets pushed toward a child.

Judge Davila emphasized this point:

The personalized-feed provisions do not prevent creators from recording or uploading content and do not prohibit direct messaging. Instead, the court explained, California adopted the provisions to protect children’s mental and physical health.

The Technology Companies Went to Court

Meta, TikTok and Google/YouTube challenged the law.

They sought preliminary injunctions potentially preventing California from enforcing the personalized-feed provisions while their constitutional challenges proceeded.

On August 5, 2026, Judge Davila rejected those requests.

The companies appealed to the Ninth Circuit.

They then returned to the district court and asked Judge Davila to prevent California from enforcing the provisions while the appeal was pending.

On August 21, he refused again.

To obtain an injunction pending appeal, the companies needed to satisfy a demanding legal standard involving, among other considerations, their likelihood of success, the possibility of irreparable injury, the balance of hardships and the public interest.

The court found their arguments insufficient.

Is an Algorithm Really “Speech”?

Perhaps the most consequential dispute involves the companies’ claim their personalized feeds constitute protected expression. This argument is important because if a platform’s recommendation algorithm is treated as the company’s own speech, government regulation of the algorithm potentially implicates the First Amendment.

But Judge Davila distinguished between the speech appearing on a platform and the predictive machinery selecting what an individual user is likely to want to see.

The court previously concluded a decision to show someone content a predictive algorithm calculating what a person will find “interesting” is non-expressive.

The algorithm, in the court’s view, essentially reflects a user’s own revealed preferences back to that user.

As Judge Davila explained, relying on a user’s characteristics and history to predict the person’s interests does not itself communicate an expressive message from the platform, This is an important distinction in the growing legal battle over social media and video game addiction.

The videos themselves may be speech. The posts may be speech. The opinions of creators unquestionably may involve speech. But the automated system analyzing a child’s behavior and deciding what content should appear next is something different.

At least according to this federal court, the First Amendment does not automatically place the choosing mechanism beyond government regulation.

The Companies Said Changing Their Platforms Would Cause Harm

The companies also argued complying with S.B. 976 would impose substantial burdens.

According to the court’s order, TikTok, Meta and YouTube raised concerns about the significant financial resources required to restructure their platforms.

Personalized feeds are central to how these services operate. The companies argued compliance could require them to develop a fundamentally different product, one users might find less appealing, costing them customers and goodwill.

Judge Davila again was unconvinced.

The court characterized the financial burdens as comparable to ordinary regulatory compliance costs, which generally do not amount to irreparable harm. The court called the risk “purely speculative.” The companies’ argument implicitly acknowledges just how central personalized feeds have become to their products.

Changing the feed, they contend, may mean changing the product itself, raising an obvious question:

Why are personalized recommendation systems so important to keeping users on these platforms?

For families involved in litigation concerning alleged social media or video game addiction, the question is likely to sound familiar.

YouTube Said Minors Could Be Harmed Without Personalized Recommendations

YouTube offered another argument.

According to the court, YouTube contended minor users could be harmed because they would lose personalized compilations identifying, among billions of videos, content likely to be “enriching, valuable, and age-appropriate.”

But S.B. 976 contains exceptions allowing minors to request media from particular creators and permitting platforms to prioritize that requested content.

Meta similarly pointed to the billions of people who use its services to share ideas.

Again, the court drew a distinction between sharing content and algorithmically delivering personalized content to minors.

S.B. 976 does not prevent those billions of users from continuing to share their ideas.

The Court then Turned to the Harm to Children

Perhaps, the most important portion of the ruling.

The court recognized California enacted the personalized-feed provisions for the purpose of protecting children’s mental and physical health.

Judge Davila described this interest as “extremely important.”

The court also referenced the U.S. Surgeon General’s Advisory cited by the statute.

According to the order, problematic and addictive social media use has been linked to:

• depression;
• anxiety;
• eating disorders; and
• other negative health outcomes.

The evidence cited by the court also stated “algorithmic designs” can promote extreme, inappropriate and harmful content to children and adolescents.

That observation goes directly to the heart of the debate surrounding social media addiction litigation.

The issue is not simply whether harmful content exists somewhere on the internet.

It is whether platforms can learn what captures a child’s attention and then repeatedly place specific content in front of that child.

There is an enormous difference between a library containing a dangerous book and a system following a child through the library, studying what catches the child’s attention, and continuously placing similar material into the child’s hands.

Modern social media recommendation systems can make billions of individualized content decisions, and the decisions are not necessarily random, they are informed by data.

The Fight Over Algorithms Is Becoming the Fight Over Product Design

For years, technology companies have frequently framed controversies involving harmful online content as disputes about speech.

But lawsuits involving children increasingly focus on something else:

product design.

The questions are changing.

Did the platform design features to maximize engagement?

Did it use behavioral information to determine what would keep a minor watching?

Did recommendation systems repeatedly deliver content based upon vulnerabilities revealed through a child’s prior behavior?

Were parents adequately informed about how these systems worked?

Were children given meaningful tools to disengage?

Could the companies have designed their products differently?

California’s law represents a legislative attempt to address some of those questions before additional harm occurs.

The litigation over S.B. 976 will help determine how much authority governments have to regulate these systems.

The Case is not Over

The companies asked a federal judge to stop California’s personalized-feed protections while their appeals proceed. The judge refused, writing:

“For these reasons, the Court DENIES Plaintiffs’ motions for an injunction of SB 976’s personalized feed provisions pending appeal.”

A Larger Reckoning Over Children’s Social Media Use

The importance of this dispute extends beyond California.

For years, parents were largely told that the responsibility for children’s social media use rested with them. Set limits, take away the phone, monitor screen time, talk to your children, etc.

But there is a fundamental imbalance.

Parents are not merely competing against a screen.

They are competing against extraordinarily sophisticated technology capable of analyzing a child’s behavior and continually adjusting what the child sees.

The companies themselves told the court personalized feeds are so central to their platforms complying with California’s law could require fundamental restructuring and potentially result in a product users find less appealing.

This admission may say more about the modern social media business model than the companies intended.

If removing or restricting personalized feeds makes a platform substantially less appealing, then perhaps we should be asking why those feeds are so effective in the first place.

Further, when the users on the receiving end of those systems are children, the question becomes much harder to dismiss.

California has decided parents should have a say before an algorithmic system delivers an addictive personalized feed to their child.

Meta, TikTok and Google are fighting this restriction.

For now, a federal court has allowed it to stand.

Litigation over social media and video game addiction continues across the country, the distinction Judge Davila drew may become increasingly important:

The right to host speech is one thing, but the right to use predictive technology to decide what keeps a child watching is another.