Should Misinformation Be Censored?

The case against censoring misinformation is powerful:

Who decides what is false? The most dangerous feature of anti-misinformation regimes is that they require someone to define truth. That power will not be exercised neutrally. Governments have obvious incentives to suppress inconvenient truths. Corporations have incentives to suppress content that threatens their interests. The history of 'official truth' enforcement is not encouraging.

False information has sometimes been official truth. During COVID-19, the lab leak hypothesis was flagged as misinformation by major platforms under pressure from public health agencies — before it received serious mainstream scientific discussion. Sweden's different approach to lockdowns was labeled dangerous misinformation — before its outcomes were analyzed. Suppressing heterodox views prevents the competition of ideas that improves collective understanding.

The legal tradition strongly protects false speech. In United States v. Alvarez (2012), the Supreme Court struck down a law criminalizing false claims about military decorations, holding that the First Amendment protects some false statements. A general power to suppress false information would be constitutionally problematic.

Counterspeech is available. The traditional answer to false speech is true speech. Fact-checking, journalism, labeling, and education can address misinformation without the dangerous power to censor.

Suppression often amplifies. Research suggests that content moderation actions can increase the visibility and credibility of the suppressed content — making the information more, not less, attractive to audiences who distrust mainstream sources.

The case for restricting misinformation argues:

Some false information causes immediate, concrete harm. Medical misinformation has contributed to vaccine hesitancy, delayed treatment, and preventable deaths. Election misinformation has disrupted democratic processes. The harm is not hypothetical.

The marketplace of ideas doesn't work when one side has overwhelming amplification. If an organized misinformation campaign can reach 50 million people and the correction reaches 50,000, the marketplace model breaks down.

Platforms already moderate. The question is not whether platforms will restrict content — they do. It is whether those restrictions should be more systematic and transparent.

Some categories of false speech are already illegal. Fraud, defamation, and perjury are all restrictions on false speech. Anti-misinformation policies are extensions of a principle already embedded in law.

Democratic self-governance depends on some shared reality. A democracy cannot function if citizens cannot agree on basic facts.

Concerns about false information and propaganda are as old as mass communication. The Alien and Sedition Acts (1798) partly aimed to prevent false claims about the government. WWI and WWII saw aggressive suppression of pacifist and defeatist speech in the name of combating enemy propaganda.

The Cold War anti-communist movement used fears about Soviet disinformation to justify domestic surveillance and speech restrictions. Each episode involved a genuine concern about false information — and each also produced significant abuses of power.

United States v. Alvarez (2012) is the most directly relevant Supreme Court case. The Court struck down the Stolen Valor Act, which criminalized false claims about military service, holding that the government cannot simply prohibit false statements without additional harm (defamation, fraud, etc.).

However, the Court left open that false speech with direct, concrete harmful effects might be regulable. The tension between this and a general anti-misinformation power remains unresolved.

AI has made misinformation both easier to create (synthetic media, automated content farms) and easier to detect (AI fact-checking, deepfake detection). It has also created a new category of concern: AI systems that are themselves trained on misinformation and may reproduce it.

Platforms' AI moderation systems for misinformation have shown high error rates — flagging accurate information as false and allowing some false information to persist. The imprecision of automated misinformation enforcement argues for careful, targeted approaches rather than sweeping suppression.