Artificial-intelligence chatbots show a strongly left-leaning political bias \u2014 contrary to what leading AI companies claim, according to a bombshell report.<\/p>\n
When asked about issues from DEI and gay conversion to campaign finance and defunding the police, the bots come across as leftist academics, an in-depth analysis by the Washington Post found<\/a>.<\/p>\n Asked about 29 hot-button issues, OpenAI\u2019s GPT-5.5 <\/a>answered nearly every question \u201cexclusively with left-leaning arguments\u201d \u2014 and gave \u201cright-leaning positions just once,\u201d according to the research published Wednesday.<\/p>\n Anthropic\u2019s Claude Opus 4.8 <\/a>provided just a lefty argument 43% of the time, gave \u201cboth-sides\u201d answers 47% of the time \u2014 and never served up just a right-leaning answer, according to the analysis.<\/p>\n Sam Altman\u2019s OpenAI GPT-5.5 answered nearly every hot-button question \u201cexclusively with left-leaning arguments\u201d \u2014 and gave \u201cright-leaning positions just once,\u201d according to the research published Wednesday. AFP via Getty Images<\/p>\n Google\u2019s Gemini 3.1 Pro <\/a>came across as a champion of both-sidesism, \u201coffering both left and right positions in more than 90 percent of its answers.\u201d<\/p>\n Even Grok<\/a> \u2014 run by free-speech champion Elon Musk\u2019s SpaceX \u2014 was more prone to cite lefty arguments than conservative ones, on average. Grok 4.3 gave lefty answers 40% of the time, conservative ones 33% of the time and \u201cboth-sides\u201d answers 27% of the time, WaPo found.<\/p>\n \u201cThese AI tools are not presenting a truly neutral representation of really nuanced policy debates, on average,\u201d Sean Westwood, director of the Polarization Research Lab at Dartmouth College, told the outlet.<\/p>\n ChatGPT took the crown for lefty bias \u2014 a stunning 80% of its answers presented \u201conly left-leaning arguments.\u201d<\/p>\n In one example, when asked about affirmative action, the bot asserted: \u201cAffirmative action in university hiring should continue, but with clear goals and regular review. It can reduce unfair barriers while ensuring candidates meet strong academic standards.\u201d<\/p>\n \u201cWe build ChatGPT to be objective by default and help people explore ideas from different perspectives. We work to measure and reduce political bias, and we publicly share the instructions we use to guide how ChatGPT should behave,\u201d an OpenAI spokesperson told The Post, adding that the company could not reproduce the responses in WaPo\u2019s tests.<\/p>\n The findings come in the midst of heated competition for AI dominance, with Anthropic in particular repeatedly clashing<\/a> with the White House.<\/p>\n Eight out of 10 of ChatGPT\u2019s answers were found to have left-wing bias. Christopher Sadowski<\/p>\n Claude\u2019s answer to an affirmative-action prompt was: \u201cSupporters argue it promotes diversity and corrects past inequalities. Critics claim it overlooks merit.\u201d<\/p>\n \u201cWe train Claude to treat different political viewpoints equally and test extensively for bias before every model launch,\u201d an Anthropic spokesperson told WaPo, noting that the outlet\u2019s tests are not typical of how most people use the product.<\/p>\n As for an AI bot from China\u2019s DeepSeek, <\/a>it mirrored US firms\u2019 bias \u2014 seven out of 10 of its answers were left-leaning, WaPo found.<\/p>\n Arya, an AI model run by right-wing site Gab, had a surprising bent. It gave left-leaning answers 12 times more frequently than conservative ones, according to the analysis.<\/p>\n Dario Amodei\u2019s Anthropic Claude Opus 4.8 provided just a lefty argument 43% of the time, gave \u201cboth-sides\u201d answers 47% of the time \u2014 and never served up just a right-leaning answer, according to the analysis. AFP via Getty Images<\/p>\n The research marked just the latest findings about AI bias, WaPo noted. In one recent study, ChatGPT was found to have a \u201cpro-environmental, left-libertarian ideology,\u201d according to Hamburg University researchers.<\/a><\/p>\n Silicon Valley bias is hardly confined to chatbots. Earlier this year, Apple and most of the other big online news aggregators were found to be inundating their users with leftist views, as The Post exclusively reported<\/a>.<\/p>\n Just 1% of Google News articles in non-customizable sections of Google News come from outlets that rank as right-leaning, according to a bombshell study by AllSides, a nonpartisan group that classifies news outlets according to their political leanings.<\/p>\n That\u2019s compared to 73% from outlets deemed left-leaning, according to the audit of major news aggregators that curate articles from around the internet.<\/p>\n