API
< Back to situation

[REVISION HISTORY]

AI influence on electoral processes

2 recorded revisions of this situation.

What changed

2026-11-07 11:02 UTC → 2026-11-13 11:42 UTC · added removed

Artificial intelligence is increasingly being integrated into electoral processes, serving both as a tool for voter assistance and a source of potential misinformation. In the United States, platforms like VoteMate have deployed chatbots such as ‘Doña Rosa’ to help voters navigate ballot measures, candidate proposals, and polling logistics in multiple languages. More recently, researchers at the University of Maryland have been testing an AI-powered chatbot named VIOLETS. Designed to provide procedural information for the 2026 midterms, the tool uses retrieval-augmented generation to pull data from official sources, though it has faced skepticism from students regarding accuracy. Simultaneously, the rise of AI has introduced risks regarding deepfakes and disinformation. In Spain, debates have emerged concerning the regulation of AI-generated satire and its potential to deceive. In the U.S., recent midterm cycles saw AI-generated political advertisements on television and social media, with approximately half utilizing the likenesses of real candidates, often without disclosing the use of AI technology. To address these challenges, OpenAI has established an Election Operations Center, described as a ‘war room,’ to monitor ChatGPT for misuse. Josh Lawson, OpenAI’s head of civic and strategic risk, stated the center aims to respond quickly to emerging issues. This move follows similar security initiatives by other technology firms, though researchers continue to highlight New research from the difficulties posed Universitat Oberta de Catalunya has highlighted inconsistencies in how AI assistants handle political queries. A study of over 1,200 queries across platforms including ChatGPT, Copilot, Gemini, Grok, Perplexity, and Google AI Overview found that responses varied by foreign-directed disinformation assistant and the limitations language, such as Spanish, Basque, Catalan, and Galician. Notably, ChatGPT provided direct voting recommendations in more than 43% of current AI guardrails. queries, while Gemini and Copilot declined to answer in over 90% of cases.

Versions

  1. 2026-11-13 11:42 UTC AI influence on electoral processes
  2. 2026-11-07 11:02 UTC AI influence on electoral processes
  3. 2026-10-28 23:16 UTC AI influence on electoral processes

Select a version to see what changed compared to the one before it.