Your membership has expired

The payment for your account couldn't be processed or you've canceled your account with us.

Re-activate

Can You Trust AI for Health Questions?

Where to look for answers in the new era of AI

person reaching for prescription bottle with other bottles nearby, sitting in front of laptop
AI systems are known to "hallucinate," generating false but plausible-sounding scientific information.
Photo: Getty Images

Querying Dr. Google is not the same as it used to be. Instead of pulling up a list of links that make it easy to identify information sources, most searches now first provide an answer or summary that’s generated by artificial intelligence.

AI uses massive amounts of data to create almost humanlike answers to questions or prompts. And at least 17 percent of adults pose health questions at least once a month to chatbots such as ChatGPT—which allow people to interact directly with AI systems—including 10 percent of people ages 65 and up, according to a 2024 survey.

AI can be useful, says Ateev Mehrotra, MD, at the Brown University School of Public Health in Providence, R.I. "But there are places where the information is inaccurate," he says, "and it’s hard for the user to know where."

In this article

3 Reasons to Be Wary

  • Fabrications can be hard to detect. AI systems are known to "hallucinate," generating false but plausible-sounding scientific information. And they can be convincing even when they’re wrong. According to a 2025 study, people often can’t tell the difference between an AI-generated answer and one from a doctor. And they often rate AI answers as better even when they’re less accurate.
  • An answer might not apply to you. An AI chatbot doesn’t know you, says Kisha Davis, MD, president-elect of the American Academy of Family Physicians, a family physician and chief health officer for Montgomery County, Maryland. It doesn’t know your specific health conditions or family history. It may generate advice based on a valid medical study, but that advice might not be relevant to your case, she points out. This is especially true for people who are underrepresented in medical research, such as women and members of minority groups.
  • Personal info may not be protected. AI systems aren’t bound by the same privacy laws as your doctor’s office. Information you provide could be used by the company that owns the system or exposed in a data breach.

How to Use AI Wisely

A 2026 study found that when provided with all the data about a particular medical case—such as a sudden, severe headache in an otherwise healthy young person—AI models were usually able to identify a relevant condition. But they performed less well when regular people chatted with the systems about those same cases because participants provided incomplete information.

More on Staying Healthy

Three steps can make a difference. First, in your prompt, tell the chatbot to rely only on information from major medical institutions, government websites, and studies published in reputable medical journals. Second, follow the links (which often appear in footnotes) to confirm whether the sources are trustworthy and the summary is accurate, Davis says. Third, talk to your doctor about what you find.

"Be skeptical," Mehrotra says. Even though AI models are improving, they can still make mistakes or expose you to risks. Double-check what you learn before acting on it.

Websites You Can Count On

If you have concerns about your health, ideally you should ask your doctor, Davis says. But if you’re looking for answers online, stick with reliable websites. These include:

Editor’s Note: This article also appeared in the May 2026 issue of Consumer Reports On Health.


Kevin Loria

Kevin Loria is a senior reporter covering health and science at Consumer Reports. He has been with CR since 2018, covering environmental health, food safety, infectious disease, fitness, and more. Previously, Kevin was a correspondent covering health, science, and the environment at Business Insider. Kevin lives in Washington, D.C., with his wife and children. Follow him on X: @kevloria.