We can trace the roots of what became the American conception of freedom of speech and the press at least as far back as the mid-seventeenth century, when John Milton, inveighing in Areopagitica against prior press licensing, described such freedom as a path to truth: if “Truth be in the field,” he wrote, “we do injuriously, by licencing and prohibiting to misdoubt her strength. Let her and Falshood grapple; who ever knew Truth put to the wors, in a free and open encounter.”

Yet truth occupies an uncomfortable place in modern First Amendment doctrine. Note the shift from Milton’s view—that freedom of speech and the press will lead the people to grasp the truth—to that of Justice Oliver Wendell Holmes, Jr. in his dissent in Abrams v. United States 275 years later (but over a century ago). For Holmes, the competition of ideas is a path to truth but not to an objective truth that is, so to speak, just out there. Holmes the pragmatist held (anti-)metaphysical views similar to those of his friend William James, defining truth in the public sphere as James did more generally, in terms of utility. When Holmes wrote in Abrams that “the best test of truth is the power of the thought to get itself accepted in the competition of the market,” he meant that what we now call the marketplace of ideas is constitutive of truth: it makes rather than reveals the truth.

Cashed out in contemporary doctrine, the Holmesian conception means that truth—in the everyday sense of objective reality—plays a very minor role. To be sure, public officials and public figures who bring defamation actions must demonstrate “actual malice,” i.e., at least reckless disregard for the truth, while other plaintiffs must prove that the reputation-damaging statement is false. More broadly, however, truth as such is mostly irrelevant to constitutional doctrine.

Speaking for a plurality in United States v. Alvarez, Justice Kennedy wrote that outside of a “few categories where the law allows content-based regulation of speech” (such as perjury and fraud), there is no “general exception to the First Amendment for false statements.” To be sure, Kennedy was not skeptical of objective truth in the same way that Holmes in the Abrams dissent was. Citing the defamation cases, he explained that there must be breathing space for the inevitable utterance of falsehoods “if there is to be an open and vigorous expression of views in public and private conversation.” Nonetheless, the upshot is a modern doctrine in which truth is largely beside the point.

That is, on balance, sensible. Defamation suits for egregious lies can still succeed, as illustrated by the $787.5 million that Fox News paid to Dominion Voting Systems to settle the latter’s claims regarding false statements in the aftermath of the 2020 presidential election. Meanwhile, a body of free speech law that made judges or juries the arbiters of truth would pose the real risks of censorship that the modern doctrine warns about.

But the fact that the Constitution is more or less neutral as between truth and falsehood creates opportunities for lies to spread. Even in Milton’s time, there was reason to doubt that truth would always or even usually prevail in a contest with falsehood. Whether people believe a proposition often has more to do with how it makes them feel, how it fits with their other beliefs, and what others in their social world believe than with the proposition’s truth value. Today, technology and political polarization exacerbate these human tendencies. Artificial intelligence facilitates the creation and distribution of seemingly real images, sounds, and videos, while the most popular platforms for sharing information are owned and operated by people and corporations with strong incentives to please a president who openly uses the levers of power to punish those who accurately report on the harm he and his administration cause.

So much for diagnosis. What is the treatment plan? I propose two interventions. First, there is a need for AI recognition tools. Google and OpenAI make it possible for at least minimally sophisticated observers to distinguish their AI-generated pictures or video from genuine content. Those are voluntary undertakings, however. Legislation mandating deployment of such technology would be a useful step and should survive constitutional scrutiny for the same reasons that mandated information disclosures (such as ingredients in food) are valid under the permissive test of Zauderer v. Office of Disc. Counsel.

Admittedly, however, such legislation would hardly be a panacea. Bad actors would likely have access to AI tools that do not comply and that could be accessed in the United States by persons using virtual private networks or generated outside the United States and then posted on platforms that reach a U.S. audience. So long as such non-compliant tools exist anywhere, it will be possible to deploy them or for people (including in the likes of the current presidential administration) to claim falsely but, for millions of credulous supporters, believably, that genuine images and videos are AI-generated.

Thus, my second proposed intervention is to empower truth-gathering and truth-verifying entities. These include traditional news organizations, but as we have witnessed recently, they are highly vulnerable to pressure, especially when embedded within larger corporations that have interests in placating the administration (e.g., CBS News) or owned by persons with such interests (e.g., The Washington Post). Mass campaigns to support standalone non-profit news organizations such as ProPublica and crowd-sourced information hubs such as Wikipedia are needed.

A recent report from Demos in the United Kingdom helpfully explicates the VDA (Verification, Deliberation, Accountability) framework for hardening democracies against the risks posed by disinformation. Its solutions are stated at a high level of generality but point the way towards some steps that can be achieved without government allies. For example, one of the eight “tracks of action” is the promotion of “Investigative Infrastructure – creating resilient, distributed systems for uncovering and verifying truth.”

One promising development along those lines is the expansion of Bellingcat, which began as a means of combining professional journalism with crowdsourced investigations of weapons in the Syrian civil war, then took on the Russian war on Ukraine, and now has global reach, including investigations and reporting in the United States. Bellingcat’s ability to provide verifiable information from war zones and inside authoritarian regimes indicates that its model—especially in its reliance on open-source material and volunteer reports subject to stringent verification procedures—should be sufficiently robust to be effective in weakened democracies such as the United States that have not yet fully deteriorated into authoritarianism.