Research Summary: Experts Disagree on How to Fight AI Disinformation, but Agree That Health and Politics Need Different Solutions

An international study involving 54 experts revealed differing perceptions of AI-generated disinformation threats across domains and significant disagreement on effective mitigation strategies. Video deepfakes were identified as the highest threat in the political domain, while AI-generated text posed the greatest concern in the health domain. Government regulation was both the most frequently cited 'most effective' and 'least effective' solution, indicating a contested landscape regarding intervention priorities.

Why it matters

This analysis highlights the complex and domain-specific nature of AI-generated disinformation threats, indicating that a one-size-fits-all approach to mitigation is unlikely to be effective. The significant disagreement among experts on regulatory efficacy underscores the need for nuanced policy development and strategic planning that accounts for varied threat vectors and perceived solution effectiveness.

Key insights

• International experts assessed AI-generated disinformation threats in political and health domains.
• Video deepfakes received the highest average threat rating in the political domain (M = 6.31/7).
• AI-generated text received the highest average threat rating in the health domain (M = 5.80).
• Experts showed divergent opinions on mitigation strategies; government regulation received both the most 'most effective' (30%) and 'least effective' (15%) votes.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.28621