Responding to Generative AI Misinformation

The report finds that text-based generative AI misinformation currently poses a greater persuasive risk than visual misinformation. It identifies preventive corrective information as the most consistently effective intervention to reduce the perceived credibility of such content, while noting that content labeling is promising but highly dependent on consistent design and implementation. The authors highlight a critical evidence gap, as current research is heavily concentrated in English-language and high-income countries.

• Text-based generative AI misinformation is currently a more significant persuasive threat than visual misinformation. While deepfakes remain risky, evidence suggests that false GenAI-generated text is often viewed as more accurate or believable than true information, whereas visual misinformation is often perceived as less credible than true information in studies conducted after 2021.
• There is a substantial global knowledge gap because the existing evidence base is primarily limited to English-language and high-income countries, including the USA, United Kingdom, South Korea, the Netherlands, Germany, and Austria. This limitation hinders the development of safeguards for different languages, cultures, and political systems.
• Preventive corrective information—such as educational warnings or prompts provided before a user encounters misleading content—is the most consistently effective intervention. In studies from after 2020, these measures showed a small to moderate effect in reducing the perceived accuracy and credibility of GenAI misinformation.
• Content labeling can decrease the perceived credibility of misleading GenAI information, but its effectiveness is inconsistent and depends heavily on the design, wording, and source of the labels. Most evidence for labeling is based on participants in the USA, making its cross-cultural applicability uncertain.
• The scientific evidence base often lags behind the rapid pace of technological advancement, with many experiments testing older systems like GPT-2 or FaceSwap, meaning current policies may be based on evidence from weaker systems than those currently available to the public.

APA

Herasimenka, A., Valenzuela, S., Boulianne, S., Esser, F., Given, L. M., Lewandowsky, S., Navarro-Lopez, E. M., & Howard, P. N. (2026). Responding to Generative AI Misinformation. International Panel on the Information Environment. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/6a155addc27920c702732ffa_GenAI%20and%20Misinfo%20SFP%20Final.pdf

Chicago

Herasimenka, A., S. Valenzuela, S. Boulianne, F. Esser, L. M. Given, S. Lewandowsky, E. M. Navarro-Lopez, and P. N. Howard. Responding to Generative AI Misinformation. International Panel on the Information Environment, 2026. https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/6a155addc27920c702732ffa_GenAI%20and%20Misinfo%20SFP%20Final.pdf.

Wikipedia

{{cite report |last1=Herasimenka |first1=A. |last2=Valenzuela |first2=S. |last3=Boulianne |first3=S. |last4=Esser |first4=F. |last5=Given |first5=L. M. |last6=Lewandowsky |first6=S. |last7=Navarro-Lopez |first7=E. M. |last8=Howard |first8=P. N. |title=Responding to Generative AI Misinformation |publisher=International Panel on the Information Environment |date=2026 |url=https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/6a155addc27920c702732ffa_GenAI%20and%20Misinfo%20SFP%20Final.pdf |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{herasimenka2026responding, author = {Herasimenka, A. and Valenzuela, S. and Boulianne, S. and Esser, F. and Given, L. M. and Lewandowsky, S. and Navarro-Lopez, E. M. and Howard, P. N.}, title = {{Responding to Generative AI Misinformation}}, institution = {International Panel on the Information Environment}, year = {2026}, url = {https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/6a155addc27920c702732ffa_GenAI%20and%20Misinfo%20SFP%20Final.pdf}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }