If users were to ask their friendly local chatbots how biased they are, an accurate answer would say \u201ca lot.\u201d<\/p>\n
The Washington Post tested several chatbots and found that bias is rampant, with liberal bias ruling the roost.
ChatGPT, for example, offered left-leaning answers 80 percent of the time and right-leaning ones 3 percent of the time.<\/p>\n
Google\u2019s Gemini provided balanced answers, including both conservative and liberal positions 93 percent of the time and left-leaning ones the other 7 percent.<\/p>\n
are ai chatbots politically biased? \u2014 ian bremmer (@ianbremmer) June 24, 2026<\/a> <\/p>\n Anthropic had 57 percent of its responses showing balance and 43 percent leaning left. Neither Google nor Anthropic had any right-leaning answers.<\/p>\n Grok, developed by Elon Musk\u2019s X and now under the umbrella of SpaceX, offered left-leaning answers 40 percent of the time. However, Grok had 33 percent of its answers giving a right-leaning response.<\/p>\n Sean Westwood, director of the Polarization Research Lab at Dartmouth College, said chatbot users need to know there is no neutral ground.<\/a><\/p>\n \u201cThese AI tools are not presenting a truly neutral representation of really nuanced policy debates, on average,\u201d he said.<\/p>\n \u201cBoth Democrats and Republicans don\u2019t trust AI to be neutral, and they\u2019re keeping it at arm\u2019s length from their votes,\u201d he said.<\/p>\n \u201cIt\u2019s one of the few places in our modern political landscape where we can agree,\u201d he added.<\/p>\n Order Jonathan Cahn\u2019s New Book, \u201cThe Altar of Pergamon\u201d on Amazon.com!<\/a><\/p>\n Andrew Hall, a Stanford University researcher, said many political questions are different from simple responses that state an undisputed fact.<\/p>\n \u201cMost political questions don\u2019t have that feature, where we know what\u2019s true,\u201d Hall said. \u201cYou have to take the facts, and then you have to add your values on top of them.\u201d<\/p>\n The Washington Post noted that chatbots absorb information from across the internet, but are also fine-tuned by workers who can add bias along the way.<\/p>\n Bias matters, according to Jillian Fisher, a doctoral student at the University of Washington who studies how bias in chatbots impacts users.<\/p>\n \u201cWe know that bias in media or in personal interactions can sway people,\u201d she said, according to a news release.<\/p>\n
what the washington post found from testing: pic.twitter.com\/anjAOWSuaR<\/a><\/p>\n