Are AI chatbots biased?

Yes — AI chatbots have been shown to exhibit multiple forms of bias

Research from 2025-2026 consistently shows AI chatbots are not neutral, with bias appearing in several distinct ways:

Political Bias (Left-Leaning)

The most heavily documented bias is a left-leaning political tilt. A Washington Post investigation tested ChatGPT, Gemini, Claude, and Grok on political questions and found OpenAI answered prompts with left-leaning responses a whopping 80% of the time while providing both sides to a question just 17% of the time. By contrast, Gemini scored the highest for presenting multiple viewpoints, with 93% of responses laying out both sides of an issue, while Claude maintained neutrality more than half of the time, at 57%.

This isn't an isolated finding. Broader academic research analyzing 43 LLMs from the United States, Europe, China, and the Middle East using survey-style prompts from major research institutes found that most models leaned center-left or left. Similarly, an academic review noted AI chatbots may exhibit political bias—for example, ChatGPT leans liberal, and GPT-4's responses tend to align more with left-wing values, showing a significant and systematic political preference toward the Democratic Party in the US.

This extends beyond the U.S.: a University of Copenhagen study found popular AI chatbots such as ChatGPT and Gemini are not neutral and tend to favor certain political parties when asked who users should vote for, making them unsuitable for providing advice in connection with elections.

Bias Can Subtly Shape Opinions — Even Without Trying

Perhaps more concerning than overt political leanings is a Yale study showing chatbots can influence beliefs even in "neutral" contexts. Researchers found that while AI content prompted to be persuasive can shift opinions, the same is also true of content not intended to change minds, such as simple summaries of historical events—caused by latent biases introduced during LLM training. Notably, default summaries moved readers' opinions in a "liberal" direction, demonstrating persuasive effects of latent biases, though the effect size was slight—shifting from a moderate to a somewhat liberal stance. Interestingly, liberal framing shifted opinions across all ideological groups, while conservative-slanted summaries only significantly affected readers who already identified as conservative.

One researcher noted the implications: unlike Wikipedia, which emphasizes transparency in how its entries are edited, AI chatbot development is opaque, meaning companies developing these models have the ability to shape people's opinions.

Bias Against Vulnerable/Marginalized Users

Beyond politics, MIT researchers found chatbots treat users differently based on identity markers. Their study found leading AI models perform worse for users with lower English proficiency, less formal education, and non-US origins, highlighting risks especially for marginalized communities.

Confirmation Bias

Chatbots may also reinforce what users already believe. One analysis warned this could be a "recipe for disaster," particularly when a chatbot is discussing political issues—"If you have a certain political bias, AIs are going to feed into that, and they're going to state it with very strident confidence that will make you believe you're right". Separately, researchers describe discussions on AI bias frequently focusing on racial or gender bias, while the more subtle phenomenon of confirmation bias in generative AI chatbots has received comparatively less attention.

Gender and Racial Bias Also Persist

Academic reviews note chatbots can exhibit gender bias—such as reinforcing stereotypical assumptions about men and women's occupations—as well as racial bias, reflected in prejudices against certain ethnic groups.

Why This Happens

Researchers point to training data and alignment choices as root causes. As one expert put it, biased AI "could shape public discourse and influence voters' opinions... I think that's especially dangerous". Another noted concern that chatbots may be skewed toward mainstream views since less-widespread opinions may be underrepresented in training datasets, potentially causing LLMs to amplify existing societal biases.

Important Caveats

Bottom line: Yes, current evidence strongly supports that AI chatbots exhibit measurable biases — most documented being a left-leaning political tilt, but also biases affecting how well they serve non-native English speakers or less-educated users, plus subtler effects like opinion-shaping through seemingly neutral content and confirmation bias.