Global Perspectives of AI Risks and Harms: Analyzing the Negative Impacts of AI Technologies as Prioritized by News Media

TL;DR

This study tackles the problem of understanding AI risks from a global, public-facing perspective by analyzing news media across 27 countries and six regions using GDELT data. It introduces an AIR-taxonomy‑based framework, augmented with 16 emergent risk categories, and combines human annotation with GPT‑4o–based classification to extract and categorize AI risk statements. The key findings show that Societal Risks and Legal & Rights‑related Risks dominate media coverage worldwide, with notable regional variation and clear influence from outlet political bias. The work provides a bottom‑up, media-informed lens for AI risk assessment that can inform more inclusive, safe, and globally attuned AI governance, policy, and regulatory planning.

Abstract

Emerging AI technologies have the potential to drive economic growth and innovation but can also pose significant risks to society. To mitigate these risks, governments, companies, and researchers have contributed regulatory frameworks, risk assessment approaches, and safety benchmarks, but these can lack nuance when considered in global deployment contexts. One way to understand these nuances is by looking at how the media reports on AI, as news media has a substantial influence on what negative impacts of AI are discussed in the public sphere and which impacts are deemed important. In this work, we analyze a broad and diverse sample of global news media spanning 27 countries across Asia, Africa, Europe, Middle East, North America, and Oceania to gain valuable insights into the risks and harms of AI technologies as reported and prioritized across media outlets in different countries. This approach reveals a skewed prioritization of Societal Risks followed by Legal & Rights-related Risks, Content Safety Risks, Cognitive Risks, Existential Risks, and Environmental Risks, as reflected in the prevalence of these risk categories in the news coverage of different nations. Furthermore, it highlights how the distribution of such concerns varies based on the political bias of news outlets, underscoring the political nature of AI risk assessment processes and public opinion. By incorporating views from various regions and political orientations for assessing the risks and harms of AI, this work presents stakeholders, such as AI developers and policy makers, with insights into the AI risks categories prioritized in the public sphere. These insights may guide the development of more inclusive, safe, and responsible AI technologies that address the diverse concerns and needs across the world.

Figures (4)

• Figure 1: Figure (A) provides an overview of the media structure of AI coverage across regions stratified by political bias. News coverage of AI in our sample is predominantly driven by news outlets with left center bias. Figure (B) shows the global prevalence of AI Risks for each category. To calculate the global prevalence, we counted the number of articles reporting risks related to each category and divided each number by the total number of articles across regions and categories (i.e., 16,312).
• Figure 2: Prevalence of AI risks in our sample from news media coverage across six different regions. The dashed red line represents the global average prevalence per risk category.
• Figure 3: Prevalence of AI risks in our sample from news media coverage across six different regions stratified by the political bias of news domains as rated by Media Bias Fact Check (MBFC) and elaborated on in Section \ref{['3.3']}. We marked the political bias leading the discourse for risks in each region by a diamond.
• Figure 4: Global prevalence of AI risks in our sample from news media coverage across the categories of political bias.