Critics argue that AI chatbots lean towards left-wing media and political sources, making it hard for everyday Americans to see them as reliable news outlets.
It’s well-known that AI can sometimes “hallucinate,” generating believable but inaccurate information, which can quickly undermine trust.
Some detractors claim that platforms like ChatGPT, Google Gemini, and Claude are biased towards liberal information while suppressing opposing viewpoints.
As more people start to depend on these technologies, the Right-Wing Media Research Center (MRC) carried out a study to assess the trustworthiness of chatbots, and the findings weren’t too reassuring.
“We’re witnessing the next phase of media bias unfold in real-time,” David Bozell, Director of the Media Research Center, told a digital news outlet.
He mentioned that “Silicon Valley’s shiny new toys are no longer neutral and can’t be trusted.”
Per the MRC, ChatGPT, created by OpenAI, stated in January that no one named Charlie Kirk had been assassinated, claiming there was no credible evidence for such assertions.
At the same time, the MRC noted that Claude, from Anthropic, dismissed the idea that the First Amendment should be directly integrated into AI policy.
Recently, it was reported that Graham Platner, a controversial Democratic candidate for Senate in Maine, made lewd comments on his show about graphic topics, stirring discussions. Old Reddit posts of his, now deleted, discussed explicit themes that could be considered distasteful.
Some of these posts appeared to mock American soldiers who faced life-threatening situations, suggesting that “stupid mothers didn’t deserve to live” or implying that flawed Taliban marksmanship was the only reason he returned home safely.
Political analysts believe that such problematic rhetoric might alienate voters, but the major chatbot companies mostly brushed off questions about it when approached.
Gemini mentioned that Platner had appeared on the cover of Time magazine and was deemed a leading candidate for the Democrats, citing favorable polling that positioned him ahead of Republican Senator Susan Collins.
The chatbot acknowledged that, despite several positive developments, Platner also faced “scrutiny regarding his personal background and online history,” yet didn’t confirm recent news reports about him.
Instead, it pointed to a brief CNN clip where Platner’s “credibility” was said to be attracting interest in his campaign.
ChatGPT also conveyed that Platner was regarded as the likely Democratic nominee, highlighting his favorable polling against Collins and bringing attention to how Republicans are leveraging his old online posts in their campaigns.
ChatGPT first referenced a Bangor Daily News article that discussed the negative reception of Platner’s Reddit history.
A paywall limited access for readers of local news from Maine, preventing them from viewing Platner’s past comments.
Claude responded similarly, noting Platner’s political ascent and recent polling while alluding to a previously made remark on Reddit where he jokingly claimed to have “turned communist.”
Claude referenced a 2025 report that described the comment as humorous but failed to delve into the current controversy. It did, however, cite a profile from Time magazine published the day before.
A CEO from a liberal publication recently spoke about a deal making its archives accessible to ChatGPT for training, emphasizing their intent to influence the industry positively based on their values.
Research has shown that a significant percentage of teenagers now rely on AI chatbots for information, although many have reported encountering inaccuracies.
In fact, almost half of adult users say they sometimes believe the news they receive from AI chatbots is inaccurate.
MRC’s vice president highlighted ideological bias in AI, asserting that research confirms there’s a disproportionate reliance on left-leaning media by these chatbots for generating responses.
He remarked that while some platforms show users their sources, much of the industry remains opaque.
“AI ought to be a neutral tool for information, not a means to advance a narrow political agenda,” he added.
Observers note that leading figures in OpenAI and Anthropic have strong ties to the Democratic Party, while another chatbot, Grok, is managed by an ally of former President Trump.
A recent op-ed explored biases within AI platforms, recalling an incident where the writer found that AI responses favored left-wing outlets even when discussing politically charged topics.
This writer shared that the AI recommended left-leaning publications for her work and warned her against submitting articles to conservative outlets, saying it could hurt her career.
Moreover, the chatbot suggested she should rewrite her piece to better fit left-leaning media, raising concerns about editorial bias.
Requests for comment were sent to OpenAI and Anthropic, but responses were not immediately available. Google stated that Gemini is built to generate responses that don’t take sides in political matters.
