The KIDS Act would put Washington in charge of how we can communicate online

Shutterstock

​This week, the House of Representatives is considering a bill to radically expand the government’s power to regulate online speech platforms. Framed as a child safety measure, the KIDS Act imposes a litany of restrictions on social media apps and other platforms.

For decades, Americans have enjoyed broad freedom to build tools to communicate with each other on the internet. From the smallest message board to the biggest social media platform, the First Amendment protects our right to freely design and use platforms to talk and share content with each other online. The government can go after wrongdoers that use online platforms for illegal activity, but it does not get to decide how a platform is designed and operated in the first place.

The KIDS Act would fundamentally overturn this status quo by allowing the government to decide how platforms can be built and imposing restrictions that will censor the speech of adults and minors alike.

Social media regulations

Title II contains the House’s version of the Kids Online Safety Act. We’ve done a deeper dive on the problems with KOSA here, but the upshot is that the bill does two major things.

First, it requires most online speech platforms to maintain policies and procedures to “address” certain harms to minors: scams, physical threats, sexual abuse, and distribution or use of drugs, tobacco, gambling, or alcohol. The exact policies and procedures platforms must maintain are not spelled out. As with the entirety of the KIDS Act, enforcement lies in the hands of the Federal Trade Commission and all 50 states, which can sue any platform they claim violates these requirements.

This sounds benign on its face, but the implications are startling: The government asserts the power to decide how we can build and run platforms for speech on the internet — in other words, how we can communicate with each other online.

Any method by which humans talk to each other — whether it’s mail, telephones, websites, or apps — will inevitably be used by some people who want to cause harm. Platforms try to detect and remove them, but when you’re sifting through thousands, millions, or billions of users, there’s no way to find all of them. The only way to guarantee people will not misuse a channel of communication is to shut it down altogether. So there is an inevitable trade-off between the freedom to communicate on a platform and the prevention of misuse of the platform.

Everyone who operates one of these platforms has to decide which security measures make sense and which would put too great a burden on users’ ability to communicate, either by imposing excessive surveillance, banning too many people, or cutting off useful features. Those choices are protected by the First Amendment.

---

---

The government, however, is now trying to seize the authority to force a different balance, to demand restrictions on speech in the name of greater security. That will burden the rights not just of minors (who enjoy robust First Amendment protections), but adults as well, since the platforms can’t know precisely which users are minors and which are adults.

The second major part of KOSA goes further in dictating a litany of specific features that anyone running a covered speech platform must build. These features must be offered to anyone whom the platform “know[s] or should have known” is a minor.

This includes things like:

• limits on the ability of other accounts to contact the user;

• limits on any design features that “result in compulsive usage” of the platform;

• an opt-out of personalized recommendation systems;

• a screen to view a list of other users who can directly message the account;

• notifications when unknown accounts try to message the user; and

• the ability to set the account to “hidden.”

These settings must be on by default for minors, and there must be a way for parents to view and control such settings.

The idea is that minors are spending too much time on their phones, and so the platforms have to offer settings that will nudge them off social media.

Again, on its face this sounds fine, but think about the implications. If the First Amendment means anything, it means we’re allowed to freely build and operate forums to communicate with each other. Yet the government claims the authority to dictate very specifically how Americans are permitted to design and operate platforms for talking to each other online. Forums large and small will now have to be designed and operated to the government’s satisfaction. Smaller platforms may be unable to even afford the costs of building all of these additional settings and keeping up with the evolving compliance requirements.

And these restrictions are built on the justification that minors are using the platforms too much — i.e., they’re consuming too much speech. If speech could be censored on the basis that too many people were choosing to listen to it, the government would have authority to restrict any mode of expression that drew the public’s attention.

The other major problem with this section is it effectively requires platforms to implement some form of age verification. We don’t know yet what it means to say a platform “should have known” a user or visitor was a minor, but we know it includes people the platform did not actually know were minors. Those users and visitors (in addition to those known to be minors) must be detected and treated differently than all the rest, and platforms that fail to do so can be sued by everyone with authority to enforce the act: the FTC and all 50 states.

As we wrote in our earlier analysis of KOSA:

This puts enormous pressure on platforms to implement age verification systems. Since age verification systems necessarily require verification of the user’s identity, their implementation would eliminate users’ ability to speak anonymously.

But the right to speak anonymously is a vital part of American history and the First Amendment. During our founding period, anonymous pamphleteers urged the colonies to break free from the English crown, and later, advocated for the adoption of the Constitution and the Bill of Rights. In the civil rights era, Alabama attempted to force the NAACP to reveal a list of its members, which the Supreme Court rightly blocked. Today, millions of Americans discuss important political and cultural topics online only because they can do so anonymously, without risking government scrutiny, professional repercussions, or social condemnation.

Beyond the problems specific to age verification, the First Amendment protects our right to access lawful speech without government conditions or hoops to jump through. Any barrier the state forces between users and protected speech puts a burden on the users’ rights. Courts generally require such barriers to be limited to measures that don’t burden our rights more than necessary to achieve the government’s goals. Here, the government quite plainly aims to suppress the consumption of lawful speech (by minors), and the bill’s sprawling regulatory regime is not narrowly tailored; it puts all manner of censorial burdens on the rights of both minors and adults.

Age verification for adult websites

This section requires platforms hosting certain kinds of adult content to implement age-verification tools.

The adult content in question is any material that meets the three-pronged Miller test for obscenity, as applied to minors. If more than one-third of the content on the platform is obscene for minors, the platform must implement a tool to detect any visitor who is more likely than not to be a minor, and then block them from seeing the content. The bill implies the FTC is to set the standards for deciding whether a visitor is a minor.

This is an effort to nationalize a version of the age verification for pornography websites that the Supreme Court upheld last year in Free Speech Coalition v. Paxton. FIRE sharply criticized the decision because it reversed decades of precedent protecting adults’ ability to speak freely online. As noted above, age verification ultimately requires identity verification for at least some portion of adult users. If an adult is flagged as 51% likely to be a minor, they’ll have to submit some kind of information that indicates their age. Although the bill says a government ID is not required, as noted above, there’s no way to verify your age without indirectly giving up your identity. Once your information has been shared, you cannot ensure it won’t be stolen or leaked.

This matters because content that is obscene for minors includes lots of speech that is fully protected by the First Amendment for adults.

There are many important things people are only willing to post or view if they can do so anonymously, and so the power to end anonymous access to that speech is effectively the power to suppress the speech itself. Recognizing this, the Supreme Court has long held the First Amendment includes protection for anonymity, including in the exact context of online pornography. In FSC v. Paxton, the court reversed itself, allowing the government to effectively suppress protected speech for adults on the grounds that it was inappropriate for minors.

Although the court’s decision was limited to age verification for sites with content that meets the high threshold of obscenity as to minors, the idea that the government can burden adults’ access to lawful speech because it’s not appropriate for kids is a Pandora’s box. Congress should rebuke it. Instead, the SCREEN Act doubles down and imposes an age verification regime even in states that have chosen to reject it and instead protect minors in other ways that don’t burden adults’ speech.

Video game regulations

Like KOSA, this section requires online video games that allow users to communicate must offer specific features to users the platform knows or should have known are minors.

The fact that the messaging features are embedded in a video game does not erode any of the First Amendment implications. Video games are protected by the First Amendment as made clear in Brown v. Entertainment Merchants Association, in which the Supreme Court struck down a California law prohibiting minors’ access to violent video games. The Court held:

Like the protected books, plays, and movies that preceded them, video games communicate ideas — and even social messages — through many familiar literary devices (such as characters, dialogue, plot, and music) and through features distinctive to the medium (such as the player’s interaction with the virtual world).

Chatbot regulations

Title IV, the SAFE BOTs Act, extends the KIDS Act’s age-based regulatory framework to AI chatbots.

For any user whom the provider of the chatbot knows or “should have known” is a minor, the bill requires chatbot providers to declare at the beginning of the interaction that the chatbot is an AI system, and then — upon receiving any user prompts about suicide and suicidal ideation — send a message with suicide and crisis hotline resources. The provider of the chatbot must also tell the user to take a break after extended interactions, and adopt unspecified policies and procedures to “address” similar harms to minors as KOSA: sexual exploitation, gambling, and the promotion of drugs, tobacco, or alcohol to minors.

AI chatbots are fundamentally platforms for speech, and so the right to freely build and use them is protected by the First Amendment. Many parts of SAFE BOTs infringe on those rights.

Like KOSA, the bill dictates to chatbot providers how they can design their platforms. No longer can someone build a chatbot with the features and messages they want to share. They must now change it to suit the government’s preferences, starting with some kind of age detection or verification on all users — eroding anonymity for everyone — and then a government-prescribed message to any user that may be a minor.

Then the chatbot must maintain procedures to “address” certain harms like “the promotion of the distribution, sale, or use” of drugs, tobacco, and alcohol. What does this mean for a chatbot? Can a state sue a chatbot provider if a chatbot shared accurate information about, say, the medical uses of opiates or the science behind alcohol distillation, on the grounds that the information might promote their use? What about producing images of historical figures smoking?

---

---

The prohibition on “stat[ing]” a chatbot is a licensed professional raises similar issues. While the government can prohibit fraudulent misrepresentations about professional licensure, the bill fails to distinguish between deception and expressive uses of AI, such as role-playing, educational simulations, or satire. Would simulations or a chatbot portraying a lawyer in an educational exercise violate the Act?

All of this speech would be protected by the First Amendment, yet the statute opens the door to government lawsuits if the FTC or a state feels it runs afoul of the bill’s regulations.

The bill’s compelled messages may appear sensible, but the First Amendment protects both the right to speak and the right not to speak. Outside the context of commercial speech like advertising, the government faces a high bar when compelling Americans to carry its messages. Yet these provisions force chatbot developers to carry the government’s speech, regardless of whether they agree with it.

Take the required notices about suicide hotlines. Even separate from questions about the accuracy of suicide-risk detection, there are sincere disagreements about the value and risks of using suicide hotlines. However, this bill puts the government in the position of deciding what messages must accompany speech, when they must be delivered, and to whom.

Child safety is an important goal. But it can never become a license for the government to dictate how Americans speak, read, and communicate online.