In describing the effects of the savage revolution that consumed the polis of Corcyra during the Peloponnesian War, the Athenian historian Thucydides laments that “words had to change their ordinary meaning and to take that which was now given to them.” Occasionally, though without the attendant bloodshed and strife, modern governments will try to transmogrify language.

For example, in almost all instances, imposing a prior restraint on speech is a grave violation of our constitutional order. The First Amendment forbids it. But the federal courts have broadly given government lots of discretion to ban price-fixing cartels. By now, that sort of thing is often treated as a perfectly acceptable, even laudable, thing for the state to do. There’s an incentive there. If you can make a prior restraint sound like an antitrust measure, maybe the government can get away with targeting protected speech.

That’s what New York State evidently thinks it can get away with. One year ago, the State purported to update its antitrust laws to preclude price-fixing and collusion between firms “with respect to residential rental dwelling units.” N.Y. Gen. Bus. Law § 340-b. Such a measure may be wise or foolish, but that kind of language sounds in antitrust.

But not in substance. New York banned—as “a class E felony punishable, for a corporation, by a fine of up to $1 million, and for a natural person, a fine of up to $100,000 and up to four years’ imprisonment”—setting an apartment’s rental rate with help from a computer program that reviews publicly available market data to generate a recommended price. Mot. for PI 8. On September 29, the Southern District of New York issued a preliminary injunction against that provision and called it by its true name. Not a valid antitrust statute, but a First Amendment violation.

The plaintiff in the case is RealPage, a company that sells AI programs covered by the State’s “antitrust” law. As relevant here, those “products generate individualized price recommendations by evaluating property-specific information provided by the user (e.g., supply and demand data, such as unit availability and apartment renewal rates) alongside publicly available rent data (e.g., market rates for similar properties).” Op. 2. While users don’t have to accept that recommendation—and many don’t—the purpose of the AI software is to allow landlords to swiftly incorporate hard data to design a proffered price. Users even “may configure a unique price-recommendation strategy by setting over a hundred custom data parameters.” Id.

It’s hard to see a valid antitrust issue here. At worst, perhaps a vast majority of real-estate firms would use the product (currently not the case) on default parameters (a bit of a leap), causing parallel pricing for all similarly situated rental units in New York. But parallel pricing, without a cartel of businesses artificially fixing the price, doesn’t harm the free market—it reflects it. The market price, after all, is the price someone’s willing to pay.

It’s very easy to see the First Amendment harm. The distribution of commercial information is squarely protected by the Constitution. And just a few Terms ago, in Moody v. NetChoice, 603 U.S. 707 (2024), the Supreme Court emphasized that the First Amendment’s embrace of speech and association doesn’t stop at the algorithm’s edge. And, as RealPage argued in its papers, the AI program’s function is to offer advice. Core First Amendment activity—from political advocacy to distributing scientific literature—can perhaps best be characterized as advice giving. By banning the use of RealPage’s algorithm, a publisher of solicited advice, New York enacted a classic prior restraint.

The district court saw the speech-eating wolf under the antitrust wool and enjoined the law. In doing so, Judge Caproni found the State’s prior restraint likely failed the exacting judicial scrutiny demanded by Central Hudson Gas & Electric Corporation v. Public Service Commission of New York, 447 U.S. 557, 564–66 (1980). That test puts the onus on the regulator to prove a measure targeting commercial speech is narrowly tailored to directly advance a substantial state interest. That means “that if the Government could achieve its interests in a manner that does not restrict speech, or that restricts less speech, the Government must do so.” Thompson v. W. States Med. Ctr., 535 U.S. 357, 371 (2002).New York flunked.

The court characterized the State’s defense as “bare bones” and noted that “an obviously less restrictive means would be to limit the prohibition to the use of software that performs a coordinating function based on non-public data from horizontal competitors.” Op. 19–20. In short, the State could pass a genuine antitrust law.  

The district court’s decision, which recognized that RealPage’s computer program is safeguarded by the First Amendment, is a welcome one. But danger does lurk. The decision comes on a preliminary posture, and the court openly noted that “the Defendant could, conceivably, persuade the Court that there is a sufficiently reasonable fit to survive exacting scrutiny under Central Hudson, particularly given the ‘considerable leeway’ afforded to the State as the executor of its governmental interests.” Id. at 20 (citation omitted). New York has already taken the court up on that offer, moving for reconsideration and pointing to an affidavit on the government’s legitimate interests already in the record. And even if that gambit fails (it should), one can expect the State to bang the table quite a bit about its discretion to ferret out anticompetitive business practices at trial or on appeal.

And it’s somewhat troubling that the district court treated this as a Central Hudson case at all. Exacting scrutiny may be a reasonable test for commercial speech, which the Supreme Court has regularly defined as speech which “does no more than propose a commercial transaction.” United States v. United Foods, Inc., 533 U.S. 405, 409 (2001). But RealPage’s advice isn’t an advertisement. It’s the kind of distribution and analysis of information that informs business practices in every sector of the economy. Surely prohibiting a company from advising a client how to act, even in a commercial context, is worthy of sterner First Amendment review than that afforded to regulating a soda commercial. (Lawyers should certainly hope so!)

This case intersects an awful lot of bad government behaviors. Absurdly aggressive antitrust theories. Prior restraints on speech. Pretending that a speech-ban is really a venerable competition measure. On this fulcrum, quite a lot of doctrine may move—so watch this space.