Toxic Synergy Between Hate Speech and Fake News Exposure

TL;DR

The paper investigates whether hate speech on Twitter correlates with exposure to fake news by analyzing the exposure of hate speakers and non-hate speakers through the accounts they follow. Using credible-source labels from NewsGuard, it shows that hate speakers encounter a higher share of low-credibility news, particularly driven by antisemitic and anti-Muslim content, and that this exposure is stronger for unpopular posts. The study also uncovers distinct political-pattern differences: antisemitic content aligns with far-left sources, while anti-Muslim content aligns with far-right sources, highlighting nuanced partisan dynamics. Although causality cannot be established, the results suggest that strategies to mitigate fake news and hate speech may have synergistic benefits for improving online discourse and safety.

Abstract

Hate speech on social media is a pressing concern. Understanding the factors associated with hate speech may help mitigate it. Here we explore the association between hate speech and exposure to fake news by studying the correlation between exposure to news from low-credibility sources through following connections and the use of hate speech on Twitter. Using news source credibility labels and a dataset of posts with hate speech targeting various populations, we find that hate speakers are exposed to lower percentages of posts linking to credible news sources. When taking the target population into account, we find that this association is mainly driven by anti-semitic and anti-Muslim content. We also observe that hate speakers are more likely to be exposed to low-credibility news with low popularity. Finally, while hate speech is associated with low-credibility news from partisan sources, we find that those sources tend to skew to the political left for antisemitic content and to the political right for hate speech targeting Muslim and Latino populations. Our results suggest that mitigating fake news and hate speech may have synergistic effects.

Figures (5)

• Figure 1: The distributions of the number of followings (left) and posts by followings (right) of hateful and non-hate speakers. The peak in the distribution of the number of followings at 5,000 is attributed to our imposed limitation, restricting the number of followings to a maximum of 5,000.
• Figure 2: Left: Distributions of the ratio of low-credibility posts to which hate and non-hate speakers are exposed. Right: Median numbers of posts linking to low-credibility and credible news sources to which hate and non-hate speakers are exposed. Error bars indicate 95% Bias-Corrected and Accelerated (BCa) confidence interval efron1987better. Significant differences in this and the following figures are indicated (***: $p<0.001$, **: $p<0.01$, *: $p<0.05$).
• Figure 3: Left: Proportions of posts linking to low-credibility news sources to which hate and non-hate speakers are exposed, for each population targeted by hate speech. Right: Median numbers of posts linking to low-credibility and credible news sources to which hate and non-hate speakers are exposed, for each target population.
• Figure 4: Median proportions of tweets linking to low-credibility news sources to which hate and non-hate speakers are exposed, broken down by popularity: less than 10 vs. more than 500 likes and retweets.
• Figure 5: Median proportions of low-credibility news to which hate and non-hate speakers are exposed, broken down by the political alignment of the source for different targets.