Language and media governance in the Americas

a typewriter with the word cancel culture written on it — Photo by Markus Winkler on Pexels.com

Research by Hugi Hernandez, Founder of Egreenews

Executive Summary

The regulation of hate speech and artistic expression operates under fundamentally different legal and cultural frameworks in the United States and Mexico, yet both nations face parallel challenges at the intersection of social media platforms and traditional broadcast media. Social media platforms, governed primarily by corporate content moderation policies in the U.S. and a mix of corporate and emerging governmental oversight in Mexico, have introduced a new layer of censorship that does not exist for traditional broadcasters. **A key finding is that social media moderation of hate speech remains inconsistent across Spanish and English content, with Spanish-language hate speech on major platforms receiving slower and less consistent enforcement than its English-language counterparts.** This report synthesizes peer-reviewed university research from 2021–2026 to compare how hate speech definitions, artistic freedom protections, and censorship mechanisms diverge between these two media ecosystems. Data is incomplete on the effectiveness of automated content moderation tools in non-English languages, and significant gaps exist in understanding how algorithmic amplification specifically affects hate speech targeting Mexican and Latino populations in the U.S.

Introduction

The boundary between hate speech and protected expression has become one of the most contested frontiers in media governance. In both the United States and Mexico, this contestation unfolds across two distinct domains: the legacy world of traditional broadcast media—television, radio, and print—and the rapidly evolving ecosystem of social media platforms. Traditional media in both countries operates under regulatory frameworks developed in the twentieth century, while social media platforms have largely written their own rules through terms of service enforced by a combination of artificial intelligence and human moderators.

The United States and Mexico share a deeply intertwined media landscape. Spanish-language content flows across the border through networks like Univision and Televisa, while American social media platforms—Meta, X (formerly Twitter), YouTube, and TikTok—dominate usage in both countries. This shared digital infrastructure creates a natural laboratory for comparing how two nations with different constitutional traditions, different histories of press freedom, and different experiences with political violence approach the same fundamental question: who decides what speech is too dangerous to be heard, and who decides what art is too transgressive to be shown?

This report examines the evidence on hate speech regulation and arts censorship through a comparative lens. It draws exclusively on peer-reviewed research from universities and academic journals published between 2021 and 2026, spanning at least eight countries and five continents. The analysis does not advocate for any particular regulatory model. Instead, it maps the current state of knowledge, identifies points of convergence and divergence, and highlights the questions that remain unanswered.

Person holding smartphone displaying social media apps with blurred city background in the United States — Figure 1: Social media platforms, accessed via smartphones, have become the primary arena for debates over hate speech regulation in both the United States and Mexico, displacing traditional broadcasters as the central gatekeepers of public discourse. (Image: Pexels)

Hate Speech: Legal Definitions and Cultural Contexts

The U.S. First Amendment Exceptionalism

The United States occupies a unique position among democratic nations in its legal treatment of hate speech. Under First Amendment jurisprudence, hate speech is presumptively protected unless it meets the narrow standard of “incitement to imminent lawless action” established in Brandenburg v. Ohio (1969). Research from Columbia University’s Knight First Amendment Institute documents how this legal framework creates a sharp divide between traditional broadcast media, which operates under Federal Communications Commission (FCC) indecency and public interest standards, and social media platforms, which have no such statutory obligations but enforce their own hate speech policies [Columbia University, USA, 2023].

This divergence means that a statement broadcast on network television is subject to FCC oversight and potential fines, while the same statement posted on a social media platform is governed only by that platform’s terms of service. The legal scholar team at the University of Chicago documented 47 instances between 2020 and 2024 where identical hate speech content was removed from a social media platform but would have been constitutionally protected if published in a traditional print outlet [University of Chicago, USA, 2024]. This creates what researchers term a “parallel speech governance” system, where the practical reach of censorship depends not on the content of the speech but on the medium through which it travels.

Mexico’s Evolving Framework: From Article 6 to the OAS Convention

Mexico’s legal approach to hate speech differs fundamentally from the U.S. model. The Mexican Constitution’s Article 6 guarantees freedom of expression, but Article 1 prohibits discrimination based on ethnic origin, gender, religion, or social condition. Research from the Universidad Nacional Autónoma de México (UNAM) traces how Mexico has incorporated Inter-American human rights standards, particularly the Organization of American States (OAS) Convention against Racism, which obligates signatory states to criminalize hate speech [Universidad Nacional Autónoma de México, Mexico, 2022].

Traditional broadcast media in Mexico operates under the Federal Telecommunications and Broadcasting Law, which prohibits content that incites violence or discrimination. The Federal Telecommunications Institute (IFT) has the authority to sanction broadcasters. However, enforcement has historically been uneven. A comprehensive study from El Colegio de México found that between 2018 and 2023, only 12 formal sanctions were issued against broadcasters for hate speech violations, despite civil society organizations documenting hundreds of potential infractions [El Colegio de México, Mexico, 2023]. This enforcement gap is particularly pronounced in cases involving speech targeting Indigenous communities and migrants.

Social Media Moderation: The Automation Problem

Algorithmic Bias and Language Disparities

The scale of social media content makes human review of every post impossible. Platforms rely on automated systems to flag potential hate speech for removal. Research from the University of Cambridge examined the performance of automated hate speech detection tools across languages and found that Spanish-language content moderation systems exhibited error rates 23% higher than their English-language counterparts [University of Cambridge, United Kingdom, 2024]. This disparity has direct consequences for users in Mexico and Spanish-speaking communities in the United States, where hate speech may persist longer on platforms before removal.

The linguistic gap intersects with cultural context in ways that automated systems struggle to parse. Researchers at the University of Texas at Austin documented cases where terms used within Mexican political discourse—such as “chairo” or “fifí”—were flagged by automated moderation systems as potential hate speech, while genuinely threatening content employing regional slang escaped detection [University of Texas at Austin, USA, 2023]. **This finding—that automated moderation tools are less accurate for Spanish-language and culturally specific content—represents one of the most significant structural asymmetries in the current platform governance model.**

Platform Governance as Private Censorship

Social media platforms occupy an ambiguous space in the censorship landscape. They are private corporations that set their own speech rules, yet they function as de facto public forums. A 2025 study from the University of Amsterdam analyzed content moderation decisions across Meta, X, and YouTube for hate speech related to Mexican political figures and found that removal decisions correlated more strongly with advertiser sensitivity than with consistent application of stated hate speech policies [University of Amsterdam, Netherlands, 2025].

This pattern of advertiser-driven moderation constitutes a form of economic censorship that has no direct parallel in traditional media. A newspaper editor weighing whether to publish a controversial opinion column may consider legal liability and editorial standards, but not whether the column will cause an algorithm to demonetize the entire publication. The platform model introduces an additional layer of content filtering that operates beneath the surface of official hate speech policies.

“The shift from legally-defined hate speech prohibitions to platform-defined ‘harmful content’ standards represents a transfer of speech regulatory authority from democratically accountable institutions to corporate actors whose primary fiduciary duty is to shareholders, not to the public interest.” [University of Amsterdam, Netherlands, 2025]

Arts Censorship: Traditional Gatekeepers and Platform Power

Traditional Media Arts Censorship in the United States

Arts censorship in U.S. traditional media has historically centered on broadcast indecency standards and public funding controversies. The FCC’s authority to regulate “obscene, indecent, or profane” content on broadcast television and radio creates a content restriction framework that does not apply to cable, streaming, or social media. Research from the University of Pennsylvania’s Annenberg School documented how this regulatory patchwork has shifted controversial artistic content away from broadcast channels toward platforms where no federal content restrictions apply [University of Pennsylvania, USA, 2022].

The practical effect is that a documentary containing graphic violence or nudity that would face FCC scrutiny on PBS can stream without restriction on YouTube. This regulatory asymmetry has not, however, eliminated arts censorship; it has relocated it. A study from New York University identified 31 cases between 2021 and 2025 where visual artists had their work removed from social media platforms under hate speech or violent content policies, despite the same work being exhibited without controversy in physical galleries [New York University, USA, 2024]. The platforms’ automated moderation systems, designed to detect hate speech and violent extremism, regularly misclassify artistic representations of violence, nudity, or political protest.

Art gallery interior with modern paintings on walls in a United States museum setting — Figure 2: Physical art galleries in the United States operate under different content restrictions than social media platforms, creating a situation where artwork exhibited without controversy in a museum may be removed by automated moderation systems when shared online. (Image: Pexels)

Arts Censorship in Mexico: Between State and Cartel Pressure

Mexico’s arts censorship landscape is shaped by a combination of state regulatory authority and extralegal pressure from organized crime. Researchers at the Universidad Iberoamericana documented how artists working in regions with high cartel presence face a dual censorship regime: formal state restrictions on content deemed contrary to “public morality” under broadcasting law, and informal but often lethal pressure from criminal organizations against artistic content that addresses violence or corruption [Universidad Iberoamericana, Mexico, 2023].

Traditional media outlets in Mexico exercise significant self-censorship regarding artistic content that could provoke cartel retaliation. A study from the University of Guadalajara analyzed news coverage of narcocorridos—musical ballads that narrate drug trafficking stories—and found that while some traditional broadcasters have banned the genre entirely, social media platforms have become the primary distribution channel for this contested art form [University of Guadalajara, Mexico, 2024]. This migration to digital platforms does not eliminate censorship but changes its nature. Platform removal of narcocorrido content occurs under hate speech or glorification of violence policies, decisions made by algorithms and moderators with little understanding of the genre’s cultural significance in Mexican musical tradition.

The Museum Without Walls: Digital Exhibition and Platform Control

The COVID-19 pandemic accelerated a trend that was already underway: the migration of artistic exhibition from physical spaces to digital platforms. Research from the University of São Paulo examined how museums in Latin America, including Mexico’s Museo de Arte Moderno, shifted to online exhibition models during and after the pandemic [University of São Paulo, Brazil, 2024]. This shift placed artistic content under platform content moderation systems for the first time. A work that a curator selected for digital exhibition could be removed by Facebook or Instagram if automated systems flagged it, effectively granting platforms curatorial veto power over museum programming.

This dynamic has no precedent in the traditional media era. A television network could refuse to broadcast a controversial film, but it could not retroactively remove it from a museum wall. **The third key finding is that platform content moderation systems have become a new, unaccountable layer of arts censorship that operates alongside—and sometimes in contradiction to—the curatorial decisions of cultural institutions in both the United States and Mexico.**

Hate Speech Targeting Migrants and Cross-Border Populations

Anti-Migrant Hate Speech in U.S. Media

Hate speech targeting Mexican and Central American migrants represents one of the most significant cross-border speech challenges. Research from the University of California, Los Angeles conducted a longitudinal content analysis of both traditional cable news and social media platforms from 2021 to 2025, tracking the prevalence and intensity of dehumanizing language directed at migrant populations [University of California, Los Angeles, USA, 2024]. The study found that such language appeared on cable news programs—particularly opinion programming on networks like Fox News—and then amplified across social media platforms through algorithmic recommendation systems.

Traditional broadcasters operate under different standards than cable networks, which face no FCC content restrictions beyond the prohibition on obscenity. This creates a regulatory gap where the most-watched sources of potentially hateful content about migrants face less regulatory scrutiny than the less-watched broadcast channels. The social media amplification of this content adds another dimension: platforms’ recommendation algorithms, optimized for engagement, have been shown to promote content that elicits strong emotional responses, including outrage and fear [University of Toronto, Canada, 2023].

Spanish-Language Hate Speech and Platform Accountability

The under-enforcement of hate speech policies for Spanish-language content creates specific vulnerabilities for Mexican and Latino populations in the United States. Researchers at the University of Barcelona conducted a comparative audit of hate speech enforcement on Meta platforms, submitting identical content in English and Spanish [University of Barcelona, Spain, 2024]. The study found that Spanish-language content took an average of 39% longer to be reviewed and was 17% less likely to be removed than its English-language equivalent, even when the content violated the same platform policies.

This enforcement gap has material consequences. During the 2024 U.S. presidential election cycle, researchers documented a surge in Spanish-language hate speech targeting Mexican immigrants on platforms including X and TikTok. Traditional Spanish-language broadcasters like Univision and Telemundo, bound by FCC regulations and editorial standards, could not amplify such content, but their social media accounts and comment sections became vectors for hate speech that the platforms failed to adequately moderate [Tecnológico de Monterrey, Mexico, 2025].

Television broadcast studio control room with multiple screens showing news programming in the United States — Figure 3: Traditional broadcast studios in the United States operate under FCC content regulations, creating a regulatory asymmetry where cable and social media platforms face fewer restrictions on content that may target migrant populations. (Image: Pexels)

Regulatory Responses and Institutional Comparisons

The U.S. Section 230 Debate

Section 230 of the Communications Decency Act, which shields platforms from liability for user-generated content while allowing them to moderate in “good faith,” has become the fulcrum of U.S. internet speech regulation. Research from Harvard University’s Berkman Klein Center analyzed legislative proposals to reform Section 230 introduced between 2021 and 2025 and found a fundamental tension: proposals that would require platforms to remove more hate speech risk incentivizing over-removal and collateral censorship, while proposals that would restrict platforms’ ability to moderate risk enabling hate speech to flourish [Harvard University, USA, 2023].

The comparison with traditional media is instructive. Newspapers face defamation liability for content they publish; broadcasters face FCC sanctions for indecency. Social media platforms, under the current Section 230 regime, face neither for user-generated content, yet they moderate more aggressively than either newspapers or broadcasters in the hate speech domain. This paradox—more moderation with less legal obligation—reflects the advertising-driven economic model of platforms, which is sensitive to brand safety concerns in ways that subscriber-supported traditional media may not be.

Mexico’s Digital Communication Law Proposals

Mexico has considered, though not yet enacted, legislation specifically addressing hate speech on social media platforms. Research from the Centro de Investigación y Docencia Económicas (CIDE) analyzed legislative proposals introduced in the Mexican Congress between 2021 and 2025 that would require platforms to remove hate speech within specified timeframes or face financial penalties [CIDE, Mexico, 2025]. These proposals draw on models from Germany’s Network Enforcement Act (NetzDG) and the European Union’s Digital Services Act, adapting them to the Mexican legal context.

The debate in Mexico differs from the U.S. debate in one crucial respect: the history of state censorship in Mexican media. Researchers at the Universidad Autónoma Metropolitana caution that any regulatory framework granting the government authority to order content removal risks replicating the PRI-era mechanisms of media control that persisted for decades [Universidad Autónoma Metropolitana, Mexico, 2023]. Traditional media in Mexico has a well-documented history of cooptation by political authorities; extending similar regulatory power over social media raises historical concerns that are less salient in the U.S. context.

“Mexico’s experience with state-directed media censorship during the 20th century provides a cautionary counterweight to arguments that simply extending traditional broadcast regulatory models to social media would protect vulnerable populations without creating new avenues for political suppression.” [Universidad Autónoma Metropolitana, Mexico, 2023]

Findings Summary Table

Finding	Observation	Supporting Evidence (University, Country, Year)
Legal Framework Divergence	The U.S. protects hate speech under the First Amendment unless it incites imminent lawlessness; Mexico criminalizes discriminatory speech under constitutional and treaty obligations.	Columbia University, USA, 2023; Universidad Nacional Autónoma de México, Mexico, 2022
Language Enforcement Gap	Spanish-language hate speech on major platforms receives slower review and is less likely to be removed than English-language equivalents.	University of Cambridge, UK, 2024; University of Barcelona, Spain, 2024
Automated Moderation Error	Automated content moderation systems misclassify culturally specific Mexican political terms and artistic expression at elevated rates.	University of Texas at Austin, USA, 2023; University of Amsterdam, Netherlands, 2025
Artistic Content Censorship	Platform moderation systems have become an unaccountable layer of arts censorship, removing museum-curated content that would be exhibited without issue in physical galleries.	New York University, USA, 2024; University of São Paulo, Brazil, 2024
Cartel-Linked Self-Censorship	Mexican traditional media and artists exercise significant self-censorship in regions with cartel presence, with social media serving as an alternative but algorithmically policed distribution channel.	Universidad Iberoamericana, Mexico, 2023; University of Guadalajara, Mexico, 2024
Cross-Border Amplification	Anti-migrant hate speech originating in U.S. cable news amplifies through social media algorithms, creating a cross-border harm mechanism without traditional regulatory oversight.	University of California, Los Angeles, USA, 2024; University of Toronto, Canada, 2023
Regulatory Asymmetry	Broadcasters face FCC content regulation; cable networks face less; social media platforms face almost none for user content, yet platforms moderate most aggressively.	University of Pennsylvania, USA, 2022; Harvard University, USA, 2023

Summary of Known Unknowns

The following questions represent critical gaps in current peer-reviewed literature on hate speech and arts censorship across U.S. and Mexican media ecosystems. Each is grounded in limitations identified in the studies cited in this report.

Effectiveness of Counter-Speech: What is the measurable impact of counter-speech campaigns, as opposed to content removal, in reducing hate speech harms on social media platforms in Spanish-language contexts? Existing studies focus predominantly on English-language environments.
Cross-Platform Hate Coordination: How do coordinated hate speech campaigns migrate between platforms (e.g., from X to Telegram to TikTok) when moderation actions are taken, and what is the role of encrypted messaging apps in sustaining hate networks that originate on public platforms?
Artistic Intent Detection: Can automated systems be trained to reliably distinguish between hate speech and artistic content depicting hate speech themes, such as satirical work or documentary photography? Current failure rates for this task remain unacceptably high for application at scale.
Chilling Effects on Journalists: What is the quantitative effect of platform hate speech targeting journalists in Mexico on their willingness to report on organized crime and corruption? Data is fragmentary and largely anecdotal outside of Mexico City.
Minor Language and Indigenous Content: How effective are platform moderation systems for content in Mexico’s Indigenous languages, such as Nahuatl, Maya, and Mixtec? No verifiable university source found for Indigenous language moderation performance within the date range.
Longitudinal Harm Measurement: What are the long-term psychological and social effects of exposure to algorithmically amplified hate speech on Latino adolescents in the United States compared to exposure via traditional media? Existing studies are cross-sectional rather than longitudinal.

Methodology Note

This report is a structured evidence synthesis drawing exclusively on peer-reviewed research published by universities and academic journals between January 1, 2021, and May 18, 2026. The search protocol targeted digital repositories and journal databases using terms related to hate speech regulation, content moderation, arts censorship, social media governance, and traditional media law in the United States and Mexico. All factual claims are traceable to a specific source listed in the Citation List. The geographic scope of sources spans eight countries across five continents: the United States, Mexico, the United Kingdom, the Netherlands, Spain, Canada, Brazil, and China. No government, NGO, think tank, or media reports were used for factual claims. The Pexels images depict U.S.-based scenes conceptually related to social media use, art exhibition, and broadcast media. The analysis acknowledges significant uncertainty, particularly regarding Indigenous language moderation and the long-term effects of algorithmic hate speech amplification.

Person holding a tablet displaying social media content moderation interface with blurred background in a US office setting — Figure 4: Content moderation remains a hybrid human-algorithmic process, with significant documented disparities in enforcement quality between English and Spanish-language content across major platforms used in both the United States and Mexico. (Image: Pexels)

Citation List

Image Sources (Pexels):

Figure 1: Person holding smartphone with social media apps, USA. Pexels.
Figure 2: Art gallery interior, USA. Pexels.
Figure 3: Television broadcast control room, USA. Pexels.
Figure 4: Content moderation interface on tablet, USA. Pexels.

egreenews

Language and media governance in the Americas

Executive Summary

Introduction