Today, a New Mexico judge ordered Meta to pay a staggering $25 billion. While most coverage has characterized it as a privacy case, the claims are fundamentally about speech, and 3 of the 5 categories of alleged misrepresentations have absolutely nothing to do with data privacy. Instead, they are focused on “hateful, violent, and divisive content,” “misinformation,” and “equal application of Facebook’s policies.” Fraud is not protected by the First Amendment, but these are incredibly vague concepts without clear legal definitions that encompass a wide array of constitutionally protected speech. A jury was asked to interpret these terms and evaluate the accuracy of Facebook’s high-level statements about how it tries to moderate content against them. To decide whether a statement like "we don't allow hate speech" was false, jurors had to apply their own understanding of terms that have no legal definition. Even more, they had to decide whether Facebook applied its policies equally. These are very amorphous concepts that mean something different to everyone, making it incredibly difficult to prove falsity, a necessary element in any fraud case. Facebook was found to have committed 43,899,725 violations. 43,899,725. How? The state multiplied the 26 statements Facebook made about its moderation practices by the number of users who may have potentially read them. Many of these statements were not advertising. They were executives speaking publicly about how Facebook handles some of the most contested content moderation questions of the last decade. The decision in this case would give government officials an incredibly powerful tool to go after content moderation decisions they don't like. Every state has a consumer protection law like New Mexico's Unfair Practices Act and every major platform publishes rules about what content it removes and how it enforces them. All a state has to do is point to an alleged discrepancy between a platform’s stated policy and its content moderation decision. If it is not overturned, this case will have a massive chilling effect on what platforms say about their content moderation practices. The safest course of action for any platform now is to say as little as possible about how it moderates content to as few people as possible. Ultimately, platforms will be less willing to share information with users and users will know less about the platforms they use.

Is it censorship when a government official asks a platform to remove content that violates its own rules? It depends! The key distinction is between asking and coercing. A request is one thing. A threat, explicit or implied, is quite another. That's a First Amendment problem.