The Spanish Language Disinformation Coalition (SLDC) defends the fundamental and equitable rights of all to express themselves, access information, and engage in the open exchange of ideas online. Yet the rise of social media influence on public discourse has also brought with it a host of complicating factors that make truly equitable free expression and civic engagement difficult. Disinformation, hateful content, activities, and other forms of information and media manipulation are particularly potent mechanisms to divide society. This coalition was created to protect against those divisive and dangerous threats posed to Spanish-speaking communities. To hold social media platforms accountable, the SLDC has developed principles following the Change the Terms model to guide platforms on how to ensure they meet their own community guidelines to prevent the spreading of disinformation and hateful activities.  Balancing the human rights principles and the concerns therein, we oppose the human and/or artificial intelligence (AI) spread, amplification, and opaque monitoring processes through which disinformation and hateful activities thrive on these platforms and set out the following guiding principles as a roadmap for platforms.

1. Social media platforms must have clear community guidelines translated into all languages in which the platform operates and make it clear that a user’s service can be terminated for engaging in hateful activities.
2. Social media platforms must prioritize preventing, curbing, and removing Spanish-language disinformation and should initiate programs and/or software trained in Spanish to counter disinformation campaigns targeting the Latinx community.
3. Social media platforms must equitably enforce their content moderation policies to ensure that the appropriate amount of attention and resources are given to moderate Spanish-language content. This includes hiring human content moderators who are fully trained on the nuances of how disinformation spreads, in addition to being Latinx and fluent in regional or local Spanish dialects and cultures.
4. Social media platforms must publicly disclose how and to what extent Spanish-language disinformation and hateful activities are proactively monitored, user-reported, and removed from the platform. Information should be included in public transparency reports that detail discrepancies between labeling or removing Spanish vs. English content and how their enforcement practices are broken down by language.

In 2020, Facebook announced a series of policy updates and enforcement actions to address militias and election misinformation in 2020. Multiple civil rights, racial justice and internet accountability organizations immediately flagged Spanish-language content that seemingly violated Facebook’s newly-updated content moderation policies. 

Violations included election misinformation, from candidate disinformation and polling/voting misinformation, and proliferation of “stop the steal” content. Other disinformation includes COVID vaccine theories and other racially charged content intended to divide the Latinx community against other races. One post in particular–a call to arms in Spanish, depicting several photos of armed white men, women, and militia groups with a caption urging white individuals to stand up and defend their lives, their flag and their country with pride–is still up and visible to Facebook users today. 

The organizations organizing this call, along with several partners, sent a series of demands to Facebook in November which have gone unmet. Read the letter here. 

For an exact timeline of NHMC’s advocacy regarding platform accountability, please visit our platform accountability timeline.

Spanish Language Disinformation Action Plan — #YaBastaFacebook!

Issued by the Center for American Progress, Free Press, National Hispanic Media Coalition and the Real Facebook Oversight Board ~ March 16, 2021

Our organizations have met with Facebook several times, we have flagged content through emails, and we have publicly shared our concerns with Facebook about the rampant spread of U.S. Spanish-language disinformation and hateful activities on the platform. We remain deeply disappointed that concerns are continuously ignored. It is time for Facebook to meaningfully act and commit to protecting its Spanish-speaking users.

We demand Facebook immediately acts on our Spanish-language disinformation action plan:

1. Hire and publicly identify a C-Suite position to oversee U.S. Spanish-language content moderation policy & enforcement.
2. Publicly explain the translation process of the content moderation algorithms, including how the algorithms are trained and audited to ensure proactive detection, reporting, and processing of Spanish-language content.
3. Clarify whether Facebook’s Content Moderators evaluate content in-language, translated, or both. Explain precisely how many US-based Spanish-language content moderators the company employs, as well as the overall number of US-based content moderators. In each case, provide the breakdown of between direct employees and contractors involved in moderation.
4. Publicly share the materials that are used to train content moderators of U.S.-based Spanish-language content.

Additional Issue Areas for Facebook’s Spanish-languge Disinformation Gap

Center for American Progress, Free Press, National Hispanic Media Coalition are founding members of Change the Terms, a set of recommendations and model corporate policies to online companies to adopt in efforts to curb hateful activities on their platforms. Facebook’s efforts to reduce hateful activities and disinformation must be replicated in Spanish and other languages in which Facebook operates. The following issue areas and action steps for Facebook align with our recommendations on matters of enforcement, training, transparency, staff, and governance.

Enforcement

• Facebook should effectively enforce the Facebook Community Standards for all bilingual and Spanish-language content at a commensurate rate of that of English content.

Evaluation and Training

• Spanish-language content moderators should be culturally competent in the social and political issues affecting the Spanish speaking community, as well as receive anti-racist training.
• Facebook should share all content moderation training materials, this includes the Known Questions and Implementation Standards documents.
• Because Spanish-language translation issues are prominent and can be delicate depending on cultural and dialect context, Facebook should regularly audit and test its machine learning algorithm for effectiveness, accuracy, and bias.

Transparency

• We demand that Facebook develop technology to begin to achieve parity in its content moderation practices across various languages, including Spanish.
• Facebook must clarify how it is training its hate speech detection AI to proactively search for content in Spanish.
• Facebook must clarify how it identifies content in Spanish for Community Guidelines or Terms of Service violations, and whether it uses its translation software for these purposes.
• Incorporate all Spanish-language content moderation data into its transparency reports.
• Include Spanish-language content moderation as an integral part to any future independent audit assessments.

Governance and Authority

• Publicly identify a C-Suite position, to overlook efforts to moderate hateful activities, misinformation, and disinformation in Spanish in the United States.
• Facebook should prioritize culturally competent candidates who are fluent in U.S. Spanish dialects for this position.