Social Media When Elections Are Close: Doing the Impossible Better
Meta has come under criticism for suspending a major political Instagram account during early voting and allegedly boosting a candidate in Brazil's recent presidential election, with the company blaming automated system errors.
With early voting in the U.S. midterms already underway, recent incidents involving Meta have highlighted the challenges social media platforms face in content moderation during critical election periods. The company suspended a major political Instagram account for hours during a significant early voting push on October 17, while also facing allegations of boosting one candidate in Brazil's recent presidential election. Meta denies deliberate interference in either case, attributing these issues to 'mistakes' in its automated content moderation systems, including AI tools that cannot distinguish between genuine follower engagement and bot attacks.
These events serve as a stark reminder that all major social media platforms need stronger systems to ensure they serve, rather than impede, democratic communication. Access to good information can profoundly affect political participation and speech, making effective moderation particularly crucial during elections.
Automation Challenges
Automated content moderation presents significant challenges beyond the recent Meta incidents. While these systems can process content at a scale impossible for humans, they reproduce existing biases, struggle with context, and make mistakes that disproportionately affect journalists, activists, artists, and marginalized communities. Language disparities exacerbate these problems, with Meta's algorithms incorrectly deleting nonviolent Arabic-language content 77% of the time in 2020, while failing to detect hate speech in many instances.
The situation is even worse for less-resourced languages. A 2025 report from the Center for Democracy and Technology found that labeled datasets in certain languages and dialects such as Maghrebi Arabic and Kiswahili contain inconsistencies, bias, and inaccuracies due to limited hiring of annotators who actually speak these languages and shifts in the languages themselves.
Systemic Failures
Beyond language issues, automated moderation systems have demonstrated systemic failures, including the suppression of content from Palestine and the repeated misclassification of LGBTQ+ content as adult or explicit material. These varied examples illustrate the risks of overreliance on automated moderation and the urgent need for stronger safeguards.
If companies are going to continue relying on automation to moderate users' speech—and there is little reason to believe they won't—then accountability must evolve alongside these technologies.
Accountability Measures
To address these challenges, experts recommend several accountability measures. Companies should publicly commit to the Santa Clara Principles 2.0, ensuring that automated technologies help rather than replace human moderators, particularly during election periods. They must also be transparent about when and how automation is used in content decisions and regularly audit their systems for bias, especially regarding low-resource languages and vulnerable communities.
Other recommendations include strengthening personal data protection policies, providing users with meaningful appeal mechanisms for moderation decisions, assessing human rights impacts of moderation practices, and carefully auditing third-party vendors. Lawmakers are advised against mandating automated moderation systems or dictating technical design choices that could favor particular expressions.
These measures recognize that content moderation is not merely a technical problem but one that requires input from policymakers, civil society, independent researchers, and affected communities. With social media platforms wielding significant power over political communication, the need for transparency, checks, and accountability has never been more urgent.
Related
