A BBC study reveals major ai chatbots frequently misrepresent news stories, raising concerns over misinformation and accuracy in AI-generated content

A recent BBC investigation has found that leading AI chatbots—including OpenAI’s ChatGPT, Microsoft’s Copilot, Google’s Gemini, and Perplexity AI—are failing to accurately summarise news stories, often introducing significant inaccuracies and distortions.

As part of its research, the BBC provided these AI models with content from its website and then assessed their responses to related questions. The findings were concerning: 51% of AI-generated answers contained major issues, while 19% of responses that cited BBC content introduced factual errors, including incorrect numbers, dates, and statements.

BBC News and Current Affairs CEO Deborah Turness expressed alarm over the findings, warning that AI developers are “playing with fire.” She cautioned that an AI-generated misinformation crisis could have real-world consequences, particularly in politically sensitive contexts. “We live in troubled times, and how long will it be before an AI-distorted headline causes significant real-world harm?” she asked.

Some notable errors highlighted in the study included:

• Google’s Gemini falsely claiming the NHS does not recommend vaping as a smoking cessation aid.
• Microsoft’s Copilot and OpenAI’s ChatGPT inaccurately stating that Rishi Sunak and Nicola Sturgeon were still in office after their departures.
• Perplexity AI misquoting a BBC report on the Middle East, misrepresenting Iran’s stance and Israel’s actions.

The investigation also found that AI models frequently failed to distinguish between opinion and fact, added editorial bias, and lacked essential context in their summaries. Microsoft’s Copilot and Google’s Gemini performed worse than OpenAI’s ChatGPT and Perplexity, according to the report.

In response to these findings, Turness urged AI companies to “pull back” their AI-generated news summaries, citing Apple’s recent decision to disable Apple Intelligence’s news summaries following similar complaints. The BBC has also called for increased collaboration between news publishers and AI developers to improve accuracy and respect editorial integrity.

An OpenAI spokesperson responded, stating: “We support publishers by helping 300 million weekly ChatGPT users discover quality content through summaries, quotes, clear links, and attribution. We continue to enhance our citation accuracy and respect publisher preferences.”

The BBC’s Programme Director for Generative AI, Pete Archer, reiterated the need for greater transparency and control, arguing that publishers “should have control over whether and how their content is used,” and calling for AI companies to disclose the scale of errors and inaccuracies in their models.

The BBC temporarily lifted its AI content block for this study, but the results underscore ongoing concerns about AI’s ability to handle news accurately. With misinformation posing an ever-growing challenge, the findings highlight the urgent need for AI developers to refine their technology before deploying it at scale.