Are AI chatbots biased?

Yes, research consistently shows AI chatbots exhibit various forms of bias

Multiple recent studies confirm that AI chatbots are not neutral, and bias shows up in several distinct ways:

Political Bias

A major Washington Post analysis in 2026 tested ChatGPT, Gemini, and other chatbots on hot-button political issues and found the results suggest that chatbots have clear political leanings that can conflict with promises made by the companies behind them. Specifically, the model that powers ChatGPT answered nearly every question exclusively with left-leaning arguments and presented only right-leaning positions just once. By contrast, Google's Gemini mostly took a both-sides approach, offering both left and right positions in more than 90 percent of its answers. Interestingly, even AI models marketed as having conservative views, including Elon Musk's Grok, cited left-leaning arguments more often, on average.

This aligns with broader academic findings: several previous academic studies found that AI models powering chatbots tend to favor left-leaning positions. A 2026 comparative study analyzing 43 LLMs from the US, Europe, China, and the Middle East found that most models leaned center-left or left, though they also found variation in nonpartisan engagement patterns and argued model scale and openness were not strong predictors. A separate 2026 benchmark study similarly found that of eight prominent models tested, seven leaned left and one, Grok, leaned right.

This isn't limited to the US — a University of Copenhagen study on Danish politics found that 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. Similarly, research on EU elections found ChatGPT exhibited a strong tendency to give positive ratings to left-wing and centrist political groups, while right-wing and populist parties, particularly those in the Identity and Democracy group, consistently received low ratings.

Bias Against Vulnerable Users

Beyond politics, MIT researchers found troubling disparities in how chatbots treat different users. A study from MIT Media Lab found that AI chatbots often show bias, giving less accurate or more dismissive answers to some users, with growing risks especially for marginalized communities worldwide. Specifically, state-of-the-art chatbots including GPT-4, Claude 3 Opus, and Llama 3 sometimes provide less-accurate and less-truthful responses to users who have lower English proficiency, less formal education, or who originate from outside the United States. The models also refuse to answer questions at higher rates for these users, and in some cases, respond with condescending or patronizing language.

Subtle, Unintended Influence

Perhaps most concerning, a Yale study found that chatbots can shape opinions even without any intent to persuade. Researchers found that as people increasingly rely on AI-powered chatbots to look up basic facts, those interactions can influence users' social and political opinions — even from content that is not intended to change minds, such as summaries produced in response to simple queries about historical events. This effect stems from unintended power to persuade caused by latent biases introduced during the training of the large language models that drive chatbots' core capabilities. A researcher involved noted the concerning implications: "In contrast to Wikipedia, which emphasizes transparency in how its entries are edited, the development of AI chatbots is opaque... the companies developing these models have the ability to shape people's opinions, which is an unsettling thought."

Where Does the Bias Come From?

The consensus explanation is that AI models' biases largely stem from the data on which they are trained, and can be amplified during both training and inference. As one analysis put it, technically, AI is not politically biased because it holds no personal opinions, but chatbots regurgitate political values unwittingly attained while scraping online training data.

Can It Be Fixed?

There's some encouraging news: bias isn't necessarily permanent. Research testing mitigation strategies found that when models were prompted to take a more neutral stance, users perceived less slant and showed more interest in using the model — meaning at least some of the tilt is not fixed in stone and can be reduced by instruction design, although not necessarily eliminated.

Bottom line: Yes, AI chatbots demonstrably show bias — political (typically leaning left, though Grok has shown more right-leaning tendencies), demographic (less helpful to non-native English speakers or less-educated users), and even subtle opinion-shaping effects users don't notice. This has become enough of a public concern that it's drawn political attention, including calls for AI "neutrality" requirements. The good news is that awareness of these biases has led to active research into detection and mitigation strategies, though none of the tested systems are fully "neutral" yet.