The biggest danger: over-censorship of legal speech
When democracies pass laws forcing platforms to remove hate speech quickly, the result is often that too much legal speech gets taken down. A 2024 study of Facebook and YouTube in France, Germany, and Sweden found that between 87.5% and 99.7% of deleted comments were actually legally permissible [1]. That means platforms are deleting mostly legal speech to avoid fines, not just illegal hate speech. This is a direct threat to free expression, and it happens even in well-established democracies.
The same study warns that such laws, like Germany's Network Enforcement Act and the EU's Digital Services Act, push platforms toward automated moderation, which is even worse at distinguishing hate speech from legitimate debate [1]. For a typical user, this means your political opinion or satire could be removed simply because an algorithm flagged it. The practical bottom line: regulation must be designed to penalize under-removal of illegal hate speech without incentivizing over-removal of legal speech.
What kind of hate speech can democracies legally ban?
Not all hate speech is the same, and democracies draw a sharp line between offensive speech and speech that directly incites violence or discrimination. The European Court of Human Rights (ECtHR) has ruled that extreme forms of hate speech—those that threaten 'tolerance, social peace, and non-discrimination'—can be excluded from free speech protection entirely [4]. For example, speech that incites genocide, as happened in Rwanda, is clearly outside democratic protection [3].
For less extreme hate speech, the ECtHR applies a proportionality test: is the restriction 'necessary in a democratic society'? They consider factors like the speaker's intent, the context, and whether the speech contributes to public debate [4]. This means a racist rant on a street corner might be banned, but a controversial academic paper on immigration would be protected. The key principle from the research: democracies can ban speech that directly causes harm (like incitement to violence), but not speech that merely offends or challenges majority views.
Why free speech still matters most in democracies
A 2025 study argues that among the classic justifications for free speech—truth, autonomy, and democracy—only the democracy argument remains fully convincing in today's world [5]. The truth argument fails because of post-truth communication and cognitive biases; the autonomy argument weakens due to social pressures. But the democracy argument holds: citizens need free speech to hold power accountable, debate public policy, and participate in self-governance [5].
This means hate speech regulation must never undermine democratic debate. The same study notes that social changes like increased individualism and migration make it harder to agree on what counts as 'harmful' speech [5]. So the practical rule is: regulation should target only speech that directly threatens democratic processes or incites violence against groups, not speech that is merely unpopular or offensive. The Rwandan genocide is a tragic example of what happens when hate speech is left unchecked—media manipulation incited mass murder [3]. But the solution is not to ban all hateful speech; it is to ban speech that directly calls for violence.
Sources used in this answer
Platform liability, hate speech and the fundamental right to free speech
A 2024 study found that 87.5% to 99.7% of deleted comments on Facebook and YouTube in France, Germany, and Sweden were legally permissible, showing platforms over-remove content to avoid regulatory penalties.
Rule of Law and Hate Speech: The Role of Criminal Law in The Realm of Free Speech (Part 1)
A 2025 analysis argues that criminal hate speech laws must respect principles of harm, proportionality, and ultima ratio (last resort) to balance free speech and protection of vulnerable groups.
From Hate Speech to Incitement to Genocide: The Role of the Media in the Rwandan Genocide
The Rwandan genocide demonstrates how media hate speech can incite genocide, proving that free speech must be limited when it directly incites human rights violations.
A Study on Hate Speech Based on Race, Ethnicity, and Religion : Focusing on the Case Law of the European Court of Human Rights
The European Court of Human Rights excludes extreme hate speech from protection under Article 17, but applies a proportionality test under Article 10(2) for less extreme cases, considering context, speaker intent, and contribution to public debate.
Free speech and social change: a public reason approach
A 2025 study concludes that only the democracy argument (not truth or autonomy) provides a public justification for free speech in contemporary societies transformed by individualism, migration, and post-truth communication.