No, the Meta Settlement Isn’t a First Amendment Problem

In the weeks since the $17.1 billion Meta consumer protection settlement, some critics have argued that the settlement raises serious concerns about—or perhaps even violates—the First Amendment. They claim that state attorneys general (AGs) engaged in a form of “jawboning”—indirectly seizing control over Meta’s application of its content moderation policies in a way that their congressional counterparts could not have achieved directly through lawmaking.

These critiques resemble a trend that Supreme Court Justice Elena Kagan has called “weaponizing the First Amendment,” which expands fringe applications of the speech clause such that the core point of the constitutional protection—fostering individual autonomy and self-governance—is diluted. Many scholars, including me, have pointed out some version of this problem.

Over the past couple of decades, US litigators arguing on behalf of tech companies and industry groups have used the First Amendment as a weapon to strike down consumer protection laws. Consider the 2011 case Sorrell v. IMS Health Inc., where the Supreme Court used the First Amendment to strike down a state law preventing drug companies from using data about the prescriptions doctors wrote in marketing designed to get doctors to prescribe different drugs. Sorrell’s innovation was to use the First Amendment to undermine consumer protection.

The novelty here is to apply the playbook to a settlement. But the idea that mutually agreed upon settlements violate the First Amendment through impermissible jawboning risks transforming conduct and products that are not speech (and certainly not speech entitled to heightened scrutiny) into speech for the purposes of the First Amendment. This reclassification is descriptively wrong, and the result is to effectively foreclose consumer protection actions that, once transformed into a speech question, will fail a heightened First Amendment test. This outcome is particularly ironic because the Meta litigation was designed precisely to avoid running afoul of the First Amendment and the third-party content immunity provided by Section 230 of the Communications Decency Act, which have been longstanding roadblocks to addressing the real social harms that platforms cause.

One reason the jawboning argument is descriptively wrong is that it assumes that Meta is an editor whose First Amendment rights might be trampled by this settlement. Maybe Facebook (as the company was known when it was founded over 20 years ago) acted like an editor when it switched from a reverse chronological timeline to moderating its News Feed, but times have changed.

In Moody v. NetChoice (2024), the Supreme Court compared Facebook’s News Feed curation to editorial judgement. Justice Barrett’s concurrence, meanwhile, states that any inquiry required to secure Moody's protections would be “fact intensive.” As such, it’s necessary to consider how much has changed since 2021, when the factual record in Moody was developed. Between then and now, Meta has drifted away from demoting certain categories of “borderline” content (for example, about COVID-19 vaccines) and leaned increasingly on personalized rankings to determine what appears in users’ feeds. This is a practice that the Moody Court described as “respond[ing] solely to how users act online—giving them the content they appear to want, without any regard to independent content standards.” And it’s a practice that the Court explicitly “d[id] not deal with.”

Meta’s abdication of editorial functions accelerated during 2025. Meta sunset its third-party fact-checking partnerships in the US in favor of Community Notes, which is driven by users. Meta doesn’t write the notes or “decide which ones show up.” It also limited what it called the “over enforcement” of its Community Standards, in part by abandoning its practice of demoting content that might have violated its rules. And, without explanation, it “removed the guidelines” for a host of content types that an editor would weigh in on, including “Content Posted by Repeat Violators of Our Policies, Posts that Indicate Suspicious Virality, and Unsafe Reporting about Suicide.” Thus, the remaining content moderation that Meta appears to engage in seems to deal with illegal or extreme forms of content—a far cry from the “wealth of user-agnostic judgments” that the Moody Court found in the company’s Community Standards and Community Guidelines.

Further, it's unclear the extent to which these judgments are being executed by artificial intelligence, which would further distance Meta from editorial discretion and closer to cold statistical probabilities. In March, Meta announced that it was shifting to using “AI systems across our apps… transforming our approach," while simultaneously reducing its "reliance on third-party vendors.” While Meta asserts that “we'll still have people who review content,” these changes run directly into the hypothetical that Justice Barrett raised that could undermine the editorial analogy: when “a platform's owners hand the reins to an AI tool and ask it simply to remove 'hateful' content," then does that technological change "attenuate the connection between content-moderation actions… and human beings' constitutionally protected right to 'decide for [themselves] the ideas and beliefs deserving of expression, consideration, and adherence.’” Indeed, one of the Court’s foundational opinions establishing “editorial control and judgment” as protected by the First Amendment, Miami Herald v. Tornillo, contemplates “the choice of material to go into a newspaper, and the decisions made as to limitations on the size and content of the paper, and treatment of public issues and public official[s],” all of which are choices made by a person.

Taken together, these moves have shifted the company away from making independent editorial decisions and toward deferring to personalized rankings and probabilistic automation. How fully Meta has abandoned Community Standards for personalized rankings and human judgment for probabilistic automation is a fact-intensive question that courts would need to answer. But the movement toward these practices, both of which de-center humans from the content moderation process, is clear. What Moody has told us is that these are the very practices that the Court excepted from its editorial analogy.

It’s notable that Meta’s policy changes to fact-checking and Community Standards were made in January 2025. If we’re going to talk about jawboning, let’s have a conversation about preemptive compliance and government coercion in the shadow of the Trump administration.

Not only does the jawboning claim fail because it is not clear that the way Meta now does content moderation qualifies for First Amendment protection under Moody, but it also fails in light of the Court’s most recent treatment of this question. In National Rifle Association of America v. Vullo (2024), the Court held that the NRA plausibly alleged that Maria Vullo, former superintendent of the New York Department of Financial Services, violated the organization’s First Amendment rights when she coerced financial institutions that her department regulated to stop doing business with the NRA and other pro-gun groups to “punish or suppress” their advocacy.

The facts in the Meta settlement are different. Vullo wielded direct regulatory and enforcement power over financial institutions, and she allegedly targeted the NRA’s speech because of its viewpoint. Here, the AGs had a cause of action that moved through the litigation process for several years and Meta—an incredibly well-resourced and sophisticated actor—had the opportunity to contest at every stage. What’s more, the AGs were not focused on Meta’s viewpoint, but rather its product design and data collection practices. (For one articulation of this argument, consider the ruling in Commonwealth v. Meta Platforms, Inc., a Massachusetts case heard earlier this year.)

Further, many of the injunctive provisions to which Meta agreed either continue existing practices, like teen content safety policies, or amount to data and design rules, like the time management provisions. Casting the settlement as violating the First Amendment risks both treating Meta as an editor for constitutional purposes and transforming its product design choices into protected speech acts. Because Meta’s changed moderation practices around personalized rankings and probabilistic automation suggest that it is no longer an editor, or at least it is a far cry from the one it once was, and because framing design as speech forecloses on legitimate products liability and consumer protections theories of accountability, those expansionist arguments should fail.

To be sure, some of the settlement’s provisions might violate the First Amendment. One of the strongest examples that critics have raised is the restriction of teen users’ ability to see ‘like’ counts on their posts. But because users aren’t a party to the agreement, they aren’t bound by it and can still file separate lawsuits over whatever First Amendment claims they may have against the states. That’s not to say it would be an easy case to make, but the claims are preserved by the settlement.

My point is not to defend all the settlement’s provisions. I share serious reservations about the age verification requirements and how they might consolidate the power of Meta’s surveillance and data collection capacities. Nor do I endorse every provisions’ efficacy. Time will tell.

The point is narrower: it’s worth critiquing the settlement, but we shouldn’t be doing it by reflexively invoking the First Amendment. Personalization and probabilistic automation are not the type of practices that we should treat as editorial judgment or expression. The Moody Court was right to flag those practices as outside of its scope of protection. And for good reason: if we extend speech coverage this way, we risk diluting this freedom by expanding its peripheral applications while eroding its core purposes of autonomy and self-governance. Further, we risk undermining the ability to hold data collection and surveillance companies like Meta accountable at a crucial inflection point where they are transforming themselves into AI companies built on the back of those practices.