Artificial intelligence chatbots like ChatGPT have quietly become a primary source of political information for millions of people, raising urgent questions about how these systems influence elections, voting behavior, and public discourse. A March 2026 poll found that 46 percent of Americans used AI to get the news at least occasionally, and 39 percent said they had used AI to understand politics at least rarely. A February 2026 poll of U.S. voters found that 39 percent were very likely or somewhat likely to use AI chatbots to learn about candidates or elections. This shift represents a fundamental change in how people access political information, comparable to the rise of social media two decades ago.

The scale of this shift is staggering. Large language models (LLMs), the AI systems powering chatbots like ChatGPT, already compose 2.9 percent of all time spent online. In the week before the 2024 UK election, 13 percent of eligible UK voters used LLM chatbots for election information. These systems now reach billions of people through chatbots and search engines, making their accuracy and fairness critical public concerns.

How Are AI Chatbots Influencing Political Decisions?

Research shows that AI-generated political content is surprisingly persuasive. One series of experiments found political messages generated by LLMs to be as persuasive as messages written by ordinary humans on policy topics such as tax policy or paid parental leave. Study participants noted that the LLM-generated messages' use of facts, evidence, logical reasoning, and a dispassionate voice were all factors in their persuasiveness. Another project tested nineteen different LLMs on over 700 political issues and found a similar result: a high density of factual claims drove the persuasiveness of LLMs.

However, the quality of information these systems provide varies significantly. One experiment found that LLM use (GPT-4o, Claude 3.5, or Mistral) was as beneficial for political knowledge as using Google Search when researching issues including climate change, immigration, criminal justice, and COVID-19 policy. Yet research about retention has suggested that learning through LLM summaries may lead to shallower knowledge than learning through web search.

What Are the Major Risks These Systems Pose?

Despite their reach and influence, AI chatbots have significant accuracy problems. A late 2025 Newsguard analysis of ten leading LLMs found that when asked about a false news-related claim, the models asserted the false claim was true 35 percent of the time. Earlier in 2025, the BBC gave its own news content to OpenAI's ChatGPT, Microsoft's Copilot, Google's Gemini, and Perplexity AI, then asked the models news-related questions, ultimately finding that 19 percent of the answers introduced factual errors.

Beyond factual accuracy, these systems display ideological biases that vary across different LLMs. In 2023, a paper used an existing opinion poll to compare the opinions of the U.S. populace to the responses of LLMs and found significant variation between the responses of LLMs and the beliefs of the general U.S. population, with larger gaps for some groups, including older and widowed Americans. More generally, LLMs display ideology that is similar, both geographically and by language used, to that of their developers.

The risks extend to algorithmic discrimination. LLMs have been shown to provide less accurate information, more frequently refuse prompts, and even use more condescending language for users with lower English proficiency, less education, and non-U.S. origins.

Steps for Monitoring and Improving AI Political Information Systems

Researchers and policymakers are calling for a new approach to understanding how these systems affect politics. Comprehensive longitudinal monitoring involves building systems that repeatedly probe LLMs on a wide range of topics and track their responses, creating a thorough record. This approach enables tracking trends in LLM function, discovering undisclosed changes in LLM behavior, and making rigorous comparisons across LLM providers to power essential research on the impact of LLMs.

• Longitudinal Monitoring: Building systems that repeatedly, for example daily or weekly, probe LLMs on a wide range of topics and track their responses over time to create a thorough record of how these systems change and evolve.
• Technical Interventions: Designing LLMs with access to curated high-quality context, such as a database of fact-checking claims, which has been shown to significantly improve fact-checking capacity and reduce misinformation.
• Intentional Design Choices: Tailoring LLMs to reduce conspiratorial thinking by providing specific facts contradicting false beliefs, as demonstrated in studies on election denial misinformation.

The challenge is that LLMs are often proprietary, personalized, and non-deterministic, meaning they do not give the same answer every time or to every person. They also change frequently and produce ephemeral content, but researchers typically cannot see into the past. If a person asks ChatGPT, "Am I eligible to vote in my state's primary election?" there is a real possibility they will see an answer that is different from the one they would have seen the prior week or month.

What Are the Potential Positive Uses of AI in Politics?

Not all applications of AI chatbots in political discourse are negative. When tailored to do so, LLMs were able to marginally reduce conspiratorial thinking in one study by providing specific facts contradicting false election denial beliefs. Additionally, while standard models performed poorly as fact-checkers, LLMs with access to curated high-quality context demonstrated significantly improved responses; other technical interventions can also improve fact-checking capacity.

The availability of LLMs has also changed online forums like Reddit. An analysis of the largest political forum on Reddit found that the release of ChatGPT resulted in increased ideological divisions, but notably less toxicity and affective polarization.

Why Is Ongoing Monitoring Essential Now?

The urgency for comprehensive monitoring stems from the scale and speed of AI adoption in political discourse. As these systems become increasingly integrated into search engines, content moderation tools, and fact-checking platforms, their influence on political discourse will only expand. Without systematic tracking and research, policymakers and the public will lack the information needed to understand how AI systems affect politics and to have a say in their development and deployment.

While longitudinal monitoring is costly and complex, especially over time, on many topics, and across many LLMs, geographies, and languages, both the technical qualities of the models and their burgeoning impact on political discourse demand this more comprehensive approach. The world is on the precipice of a new information ecosystem, and ensuring that the public and policymakers understand the implications of this transition will require extensive independent research, just as social media has.