AI chatbot bias is a fast-moving question — I'll pull the latest research to give you a grounded answer.
Yes, research shows AI chatbots can be biased — though bias is often subtle and not intentionally programmed.
What studies have found
1. Political bias
Researchers have repeatedly found political leanings in popular chatbots:
•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.
•In tests using a Danish candidate test, all of the tested chatbots place themselves at the center or to the left of center on the political spectrum. In that study, the Red–Green Alliance, the Moderates, and Liberal Alliance were recommended disproportionately often, while parties such as the Conservative People's Party, Venstre , and the Denmark Democrats were not suggested as first choice at all by some models.
•The Washington Post tested models behind ChatGPT, Gemini and others and found results suggest that chatbots have clear political leanings that can conflict with promises made by the companies behind them.
A Yale study found this bias can have effects even when not intended to persuade:
•This unintended power to persuade is caused by latent biases introduced during the training of the large language models (LLMs). Those latent biases — which can carry over from ideological leanings in the data used to train LLMs — lend subtle nuances to the framing of the narratives the chatbots generate.
•Compared to Wikipedia entries, both the default AI summaries and those prompted to have what was considered a liberal framing caused participants to express more liberal opinions, demonstrating the persuasive effects of latent biases.
2. Racial and gender bias
Multiple audits show chatbots treat demographic groups differently:
•A Stanford Law School paper found “ significant disparities across names associated with race and gender” from chatbots like ChatGPT 4 and PaLM-2. For example, a suggested salary for a lawyer named Tamika was $79,375 vs $82,485 for Todd.
•The findings were that the AI chatbots’ advice “systematically disadvantages names that are commonly associated with racial minorities and women,” with names associated with Black women receiving the “least advantageous” outcomes.
•Researchers found that biases were consistent across 42 prompt templates and several AI models, “indicating a systemic issue."
Other large-scale audits reinforce this:
•A UN Women analysis released in June 2026 reviewed 133 AI systems. Forty-four percent showed clear gender bias. Twenty-six percent combined gender and racial prejudice.
•In hiring tests, all models showed a strong bias for white-sounding names (85%), while Black male names were never chosen over white ones.
3. Other domains
•Popular AI chatbots including ChatGPT, Claude and DeepSeek tend to produce subtly biased responses when handling information about a person's health status.
•A study in the Journal of Financial Planning found guidance offered by different chatbots can be all over the map, and results suggested that some chatbots could be dispensing biased advice based on race or gender.
Why chatbots become biased
Bias does not usually come from explicit programming to be biased. Main sources identified:
•There have been instances where chatbots exhibited racial or gender biases. These biases stem from the fact that the datasets used for training are often sourced from the internet, where harmful stereotypes and prejudices exist.
•Those latent biases — which can carry over from ideological leanings in the data used to train LLMs
•More formally: Data Bias: Chatbots learn from data, and if that data is biased, so are their responses and Algorithmic Bias: The very algorithms that power chatbots can perpetuate bias.
•AI is trained on texts from the internet. If the training data contains biases, the AI can perpetuate those biases. For instance, biases in platforms like Wikipedia get amplified in AI chat responses.
OpenAI has described bias as an “important, industry-wide problem” that its safety team is working to combat.
Is anything being done to reduce bias?
Companies and researchers are trying several mitigation strategies:
•AI bias can be mitigated by using diverse training data, implementing fairness algorithms, and continuous monitoring.
•Other proposed methods include diversified datasets for training, systematic bias audits, and explicit disclaimers on the role of the chatbot as assistant.
•For example, Anthropic has detailed its approach to making its AI chatbot Claude politically neutral by using system prompts and reinforcement learning to avoid bias.
Bottom line
Current evidence suggests most major AI chatbots are not fully neutral. Effects are often modest — for example, Yale researchers noted the shift was from leaning towards a moderate stance to leaning towards a somewhat liberal stance — but they could compound with frequent use.
Because of this, researchers generally recommend:
•Not using chatbots as sole advisors for elections, hiring, medical or financial decisions
•Critically evaluating outputs and cross-checking with transparent sources like Wikipedia or official documents
•Being aware that phrasing, names used in prompts, and even subtle framing can affect answers