On September 28th, California Governor Jerry Brown signed a new law that forbids the surreptitious use of bots to, among other things, influence Californians’ votes in an election. The law is a disclosure rule: the bot may speak, but it must clearly identify itself as nonhuman.

Critics have suggested that California’s new law is vague and unworkable and will stifle legitimate speech. The First Amendment looms large in this debate.

But before rushing to First Amendment judgment about a law regulating speech, it is worth pausing to consider the gravity of the risks the law seeks to contain, and to ask if a commitment to free speech precludes efforts to address those risks. California’s law stems from growing alarm about our ability to tell what is real from what is fake when we talk politics. As computers become ever-more sophisticated at producing them, we face a reality in which our marketplace of ideas is increasingly flooded with counterfeits.

And bots are the least of our problems. Cheap and publicly available software will very soon be capable of producing hyper-realistic videos of candidates for office saying things they did not in fact say. Shouldn’t a state be able to prohibit this deception just as it prohibits counterfeit cash from circulating?

Plenty of laws confront the problem of counterfeits outside the political realm. We have a long history of policing them. It is illegal to circulate fake money. Sellers of fake goods – even or especially when they are of the same quality as originals – violate consumer protection and trademark laws. Criminal laws punish posing as a police officer to get out of a speeding ticket. The urge to prohibit fraud in these contexts makes sense.

In a forthcoming article, I explore whether a ban “counterfeit campaign speech” is constitutionally possible. A ban on counterfeit campaign speech would address the manufacture of fake images or audio of an identifiable candidate for public office to create a false appearance that the candidate has done or said something that he or she has not done or said. It would prohibit only fakes produced with knowing or reckless disregard of fake-ness and an intent to harm a candidate’s chance of electoral success, mislead voters, and undermine the electoral process.

The problem with such a prohibition is that courts have consistently held that, though we can ban frauds and counterfeits in other contexts, elections are different. When it comes to political speech, the First Amendment severely and appropriately curtails efforts to combat deception. Court after court has ruled that attempts to prevent lies in political campaigns chill speech and put the state in the impossible position of discerning truth from falsity. Far better to let the marketplace of ideas sort out what is true and what is not when it comes to politics.

Banning counterfeit campaign speech therefore feels like a radical proposition. But a careful reading of cases like New York Times v. Sullivan reveals that the Supreme Court does not extend First Amendment immunity to calculated falsehood that wreaks specific and identifiable harm. Assuming that a prohibition of counterfeit campaign speech reaches only fraudulent speech that is knowingly so, no First Amendment protections should attach. This argument has not won the day when it comes to regular lies. But faked speech is different than a regular lie—it’s fraud. And scholars who have examined laws banning lies in political campaigns have left the door open to narrow prohibition. A prohibition of faked candidate speech could fit that bill.

California’s bot disclosure statute may skirt First Amendment scrutiny since it does not ban or penalize speech; it requires only that bots disclose their non-humanity. Thus far, the Supreme Court has supported disclosure requirements against First Amendment challenges in a variety of campaigning contexts. A broad law banning bot speech outright would be unlikely to pass a First Amendment test.

But the same should not be so for a law imposing individual liability for intentionally faked candidate speech—at issue is not the means by which a message is conveyed but whether it is an intentional fake.

The trouble is that even if a ban on counterfeit campaign speech survived constitutional scrutiny, a prohibition would still face numerous practical hurdles. Identifying the source of faked speech or even that it is fake is tricky business (though not impossible). It is not at all clear that the government is capable of policing counterfeit campaign speech even if a law were on the books. Critics have expressed similar reservations about the enforcement of California’s bot disclosure law, for example the risks it poses to unmasking anonymous speakers.

Given this, maybe a ban on counterfeit campaign speech is futile, especially amidst the deluge of misinformation already drenching voters. But why not flip the narrative? Why not assume that protecting voters from counterfeits is even more important than protecting consumers from fake Nikes? Our democracy may depend on it.

Read more at Harvard Law Review