Does AI Fight or Fuel Election Disinformation?

By Brennan Center for Justice.

Summary: Examines evidence on whether generative AI is amplifying or helping counter election-related disinformation.

Relevance: Relevant to Surveillance & Civic Life: tests how the chatbots clients and communities increasingly consult for political information (ChatGPT, Gemini, Grok, Claude, Perplexity, DeepSeek) perform on election disinformation, bearing on digital literacy work with clients.

Key findings: All six chatbots consistently disputed known conspiracy theories (noncitizen voting, rigged machines) even under repeated questioning, but every tool readily generated realistic fake images of voter fraud on request, and four of the six correctly flagged fewer than 4 of 14 AI-generated test images as fake. Brennan Center found roughly half of chatbot responses to election questions contained inaccuracies or bad citations.

Read the original: https://www.brennancenter.org/our-work/research-reports/does-ai-fight-or-fuel-election-disinformation