From hate speech to toxicity: depoliticization, algorithmic governance, and the transformation of harm in digital public discourse

Over the last twenty-five years, hate speech has become a key category in international public policies, while digital environments have increasingly promoted the rise of the broader and more operational category of toxicity. This article argues that the shift from hate speech to toxic content should not be understood as a merely terminological substitution, but as a semantic and governmental transformation in the way discursive harm is identified, measured, and managed. The paper first reconstructs the historical and normative genealogy of hate speech, then examines the psychological, computational, and platform-based genealogy of toxicity, and finally compares the two frameworks through their conceptual, operational, and political implications. Particular attention is paid to the agency of platforms, algorithmic governance, content moderation, and the tension between discriminatory harm and conversational harm. The article suggests that toxicity offers scalability and technical operability, but may also contribute to the depoliticization of online harm if detached from histories of discrimination, protected characteristics, and asymmetries of power.