Consumer Reports advises treating chatbot responses as a starting point rather than diagnosis or treatment. Survey and study findings cited in the report show regular use, overconfidence and difficulty distinguishing AI answers from clinicians’ responses.
Consumer Reports said AI health answers can fabricate facts or mix up important details. A survey cited in the report found 17% of adults ask AI chatbots health questions at least monthly. A separate study found many respondents could not reliably distinguish AI-generated answers from a doctor’s.
Consumer chatbots are not governed by the same privacy relationship as a doctor’s office. The report advised users not to enter sensitive personal health information. It recommended verifying information with sources such as MedlinePlus, CDC or recognized medical organizations and calling 911 for emergencies.
General-purpose chatbots generate probable text rather than performing a clinical examination. Medical decisions depend on history, testing, dosage and urgent warning signs that may be absent from a prompt. Privacy terms vary by service and can change over time.
At the publication cutoff, the checked record establishes the following limit: The cited survey does not measure every product or medical specialty, and accuracy varies by question and model version. The next public records expected to update this account are peer-reviewed evaluations of current models and new privacy controls and clinical oversight requirements.
This article separates completed events from announcements, allegations, forecasts and proposed actions. Statements by governments, companies, litigants and other interested parties establish what those parties said; they are not treated as independent proof of every underlying assertion. Numerical values describe the source period and category identified in the reporting and may be revised by the responsible institution.
The account is fixed to the August 1 edition cutoff and retains the exact source pages checked for Health Chatbot Use Raises Accuracy and Privacy Warnings. A later filing, official release, verified field report or corrected dataset could change a total, timetable or attribution without erasing the documented sequence available at this cutoff. No image is included because no rights-cleared visual was required to report the facts.