Guardian Criticises Google AI Health Advice

An investigation by The Guardian has claimed that Google’s AI Overviews provided misleading or inaccurate information in response to certain health-related searches. Google has challenged the findings, arguing that many of the examples reviewed were taken from incomplete screenshots and do not reflect the overall accuracy of the feature.

According to the report, The Guardian tested a selection of medical searches and asked health charities, clinicians, and patient information organisations to assess the AI-generated summaries shown at the top of Google search results. Several experts involved in the review said some responses contained guidance they believed was incorrect or potentially harmful.

Health Experts Highlight Specific Concerns

One of the examples cited involved pancreatic cancer. Anna Jewell, director of support, research and influencing at Pancreatic Cancer UK, said advice suggesting patients avoid high-fat foods was inaccurate. She warned that following such guidance could seriously affect a patient’s ability to remain well enough for treatment.

Mental health queries were also flagged. Stephen Buckley, head of information at Mind, said some AI-generated summaries relating to conditions such as psychosis and eating disorders included advice that was dangerous, misleading, or might discourage people from seeking professional help.

The investigation also referenced cancer screening information. Athena Lamnisos, chief executive of The Eve Appeal, said an AI Overview incorrectly listed a cervical smear test as a screening method for vaginal cancer, describing the information as entirely wrong.

Sophie Randall, director of the Patient Information Forum, said the examples showed how AI Overviews can elevate inaccurate health information to the top of search results, potentially placing users at risk.

The Guardian also noted that running the same health search at different times could generate different AI summaries, drawing on varying sources and producing inconsistent results.

Google Pushes Back on the Findings

Google disputed both the conclusions and the examples cited. A spokesperson told The Guardian that many of the screenshots shared were incomplete, but that the AI Overviews generally link to trusted, well-known sources and encourage users to seek expert advice.

The company said the vast majority of AI Overviews are accurate and helpful, adding that it continues to refine the system. Google also stated that the accuracy of AI Overviews is comparable to other search features, including featured snippets.

Google acknowledged that errors can occur if context is missed or content is misinterpreted, and said it takes corrective action in line with its policies when issues are identified.

Part of a Wider Debate on AI Accuracy

The report arrives amid ongoing scrutiny of AI Overviews, which expanded more widely in 2024. Early in the rollout, the feature attracted attention for producing odd or incorrect answers, prompting Google to narrow the types of queries that trigger AI-generated summaries and to improve safeguards.

Since then, concerns have continued, particularly around health-related searches. Research from Ahrefs suggests that medical “Your Money or Your Life” queries are more likely than average to display AI Overviews, appearing in over 44% of such searches analysed.

Separate studies examining large language models in medical contexts have also raised questions about citation quality, finding that some AI-generated answers are not fully supported by the sources they reference.

Why the Issue Matters

AI Overviews appear above traditional search results, giving them significant visibility. When the subject is health, even small inaccuracies can have serious consequences.

Publishers and health organisations have long been required to meet strict standards around expertise and trust. The Guardian’s investigation highlights growing attention on whether Google’s own AI-generated summaries should be held to similar expectations.

The variability of AI responses over time also presents challenges, making it difficult for users, regulators, or researchers to reliably review or verify what information was shown.

What Happens Next

Google has previously made adjustments to AI Overviews following public criticism. Its response to The Guardian suggests it sees the feature as an extension of existing search tools rather than something that should be judged by a different standard.

As scrutiny continues, the debate is likely to focus on how AI-generated health information is tested, reviewed, and corrected, particularly when it appears at the very top of search results.