At this point, chatbot like Chat GPT or gemini. In addition to general questions, summarize long textsyou may have even had health-related questions. Perhaps you were trying to decide whether the symptoms were cause for concern or trying to understand test results in the middle of the night when your doctor wasn’t around.
According to OpenAI’s January 2026 report, more than 5% of all ChatGPT messages worldwide are related to healthcare, and more than 40 million weekly active users around the world ask healthcare questions every day.
(Disclosure: CNET’s parent company, Ziff Davis, filed a lawsuit against OpenAI in 2025, alleging that it infringed on Ziff Davis’ copyrights in training and operating AI systems.)
In January 2026, OpenAI introduced ChatGPT Health, which turns habits into a purpose-built feature. This “health-focused experience” within ChatGPT is designed to help you understand medical information and prepare for a real-life conversation with a clinician.
It is not a diagnostic or therapeutic tool.
So what exactly is ChatGPT Health and how is it different from Ask a Question? chatbot Random health question? Let’s double-click on ChatGPT’s new Health tab.
What is ChatGPT Health?
Chat GPT Health Not a separate app as such AI browserAtlas. It exists as a dedicated space or tab within ChatGPT and focuses on health-related questions, documents, and workflows.
ChatGPT[ヘルス]tab.
OpenAI said in its release notes that it worked with more than 260 doctors practicing in 60 countries and dozens of specialties over two years and reviewed health-related model responses more than 600,000 times. As a result, we do not respond to medical content in the same free-form format as regular chat. Instead, we will be more cautious by more restricting how information is presented and by more clearly encouraging people to seek professional care.
ChatGPT Health is available on the web and mobile app. You don’t need to download anything or sign up outside of ChatGPT itself. Access varies by location and rollout stage, but you can join a waiting list.
As of early 2026, ChatGPT is available in the United States, Canada, Australia, parts of Asia, and Latin America, where it already supports health features. Currently not available in the EU, European Economic Area, UK, China, or Russia. OpenAI says availability will expand, but timelines will vary by region due to local regulations and health data rules.
How ChatGPT Health works
ChatGPT Health uses the same foundation Large-scale language model (LLM) As ChatGPT. You ask a question, and the model generates a response. The difference with Health is context, rationale, and constraints.
Alex Kotlar, founder of Bystro AI, a genetics-focused LLM platform that provides health insights, told CNET that OpenAI is not building a new fundamental health model.
“They haven’t created a model that suddenly understands medical records better. It’s still ChatGPT, it’s just connected to medical records,” Kotlar said.
Health comes from the data you choose to sync, but it cannot be accessed unless you explicitly grant permission. In addition to medical records, you can connect other apps such as: apple healthlab results from Function, and meal logs from my fitness pal. You can also link weight watcher for GLP-1 Instacart, which turns meal ideas and meal plans into shopping lists; peloton When it comes to training recommendations, this allows the AI to provide personalized insights based on history rather than general advice.
These conversations now follow stricter rules regarding tone, source, and response style, and OpenAI says it is testing them using the assessment framework HealthBench. We use physician-written rubrics to evaluate model responses across 5,000 simulated health conversations and apply over 48,000 specific criteria to assess quality and safety.
You can upload documents to Healthcare, use voice commands, and do everything you would do in regular chat. If you’re reviewing multiple test results or preparing to see a specialist, Health can help you keep track of what you’ve already shared and organize your information over time.
OpenAI suggests using it to review test results, organize pre-appointment questions, translate medical terminology into plain English, and summarize long documents such as hospital discharge records or insurance instructions.
OpenAI makes clear that the tool is intended to assist in conversations with medical professionals, not to shortcut them. They cannot order tests, prescribe medications, or confirm diagnoses. If you treat it like a doctor, you are using it incorrectly.
But Dr. Saurabh Gomber, a clinical instructor at Stanford Health Care and chief medical officer of Atropos Health, told CNET, “I think the preparation and education itself is actually already beyond the boundaries of medical advice.”
Health separates conversations, connected apps, files, and Health-specific memory from the main chat, so details about your health don’t leak back into the rest of ChatGPT. However, health can use normal chat memory. Let’s say you mentioned a recent move or lifestyle change (like going vegan). Health can use that context to make conversations more relevant.
You can also track patterns over time. If you connect apple healthyou can ask about sleep See trends, activity patterns, and other indicators, and use that summary to talk to your doctor.
ChatGPT Health and ChatGPT for Healthcare are not the same
ChatGPT Health is a consumer feature for personal health. According to OpenAI’s Help Center, HIPAA does not apply to consumer health products such as Health. HIPAA is the Health Insurance Portability and Accountability Act, which provides federal protections against disclosure of sensitive patient health information without consent.
Separately, OpenAI offers “ChatGPT for Healthcare” for organizations that require controls designed for regulated clinical use and support for HIPAA compliance, including business associate agreements (such as agreements between healthcare providers and billing companies).
If you use ChatGPT Health, you can connect your medical records to this feature, but they are not part of your hospital system. We use consumer products with additional protections controlled by OpenAI. Therefore, do not assume that “health features” automatically equals HIPAA.
Privacy and data management
According to OpenAI, Health adds additional protections on top of ChatGPT’s existing controls, including “proprietary encryption and isolation to secure and compartmentalize health conversations.”
An OpenAI spokesperson told CNET that ChatGPT conversations and files are encrypted by default at rest and in transit, adding an additional layer of protection with the sensitivity of health data in mind. The spokesperson added that employee access to health data is limited to safety and security duties, and that access is more restricted and purposeful than normal product data flows.
“When consumers hear that something is encrypted, they often think that no one can see it, but that’s not the case. Just because it’s encrypted at rest doesn’t mean the company itself can’t access the data,” Kotlar said.
You can disconnect apps, remove access to your medical records, and delete health memories. Dane Stuckey, chief information security officer at OpenAI, also said that health conversations are not used to train the underlying model by default.
Still, “more protected than regular chat” doesn’t mean it’s risk-free. Even strong security cannot eliminate everything risk Relates to storing sensitive health information online. That’s one of the reasons for privacy Experts urged users to Think carefully before uploading your complete medical record to an AI tool.
Limitations and safety concerns
ChatGPT Health helps you understand the information, but things could still go wrong. There is a risk that people will take health care answers at face value. In today’s world, everyone wants answers right away, and AI sounds convincing even when it should be cautious. Unsurprisingly, ECRI (a patient safety nonprofit organization) has named AI chatbots as the No. 1 health technology standard for 2026.
hallucinationan AI’s habit of confidently generating incorrect details is far more important in medicine than telling it to do so. Summarize PDF. If a tool completely fabricates research, misreads test values, or exaggerates the meaning of symptoms, it can seriously jeopardize your health.
“The biggest danger for consumers is that unless they have medical knowledge, they have a hard time determining when they’re saying the right thing and when they’re saying the wrong thing,” Kotlar said.
When asked for comment on the hallucination rate, an OpenAI spokesperson said the model powered by ChatGPT Health has “dramatically reduced” hallucinations and other high-risk errors in difficult medical conversations. According to OpenAI’s internal HealthBench assessment: GPT-5 Compared to previous models, it reduces hallucinations in difficult health scenarios by 8x and errors in potentially emergency situations by more than 50x. GPT-4o It also shows that there are no undetected failures in adjusting to the global health situation.
The company also claims that the new model is significantly more likely to ask follow-up questions in cases of uncertainty, which reduces the risk of giving confident but inaccurate answers.
“Companies that are using these tools need to share that they have these kinds of checks and balances and benchmarks in place to ensure that the quality of the answers is high,” Gomber told CNET.
OpenAI argues that Health is not intended to be diagnostic or therapeutic and should support, rather than replace, clinician care.
Kotlar said such tools are “both great and kind of irresponsible” because healthcare is complex and highly regulated.
Should I use it?
People were already using ChatGPT for health-related questions, even when it wasn’t the right place to ask health-related questions. Health adds stricter guardrails for medical topics, so it may be slightly safer than standard chat for that purpose. According to the American Medical Association, which surveyed more than 1,100 physicians, physicians are increasingly using AI in their practice, with usage nearly doubling from 2023 to 2024.
Gomber added that while the interaction between traditional medicine and AI is likely to continue, “there is room for improvement and benefit.” many people no reliable access And if an LLM can at least help someone triage whether or not they should see a doctor, that alone might make sense.
If you’re careful about the data you share, it can be helpful. Use Health to translate medical terminology, draft questions, and more so you don’t freeze up during your 10-minute appointment. You can also ask other low-risk questions about diet and exercise. Always verify information with reliable sources or experts. It should never be used for self-diagnosis, deciding to take or stop medication, or interpreting serious symptoms.
“Medicine is different than coding or writing,” Kotler said. “When we fail, we fail in ways that are really dangerous to humans.”
ChatGPT Health helps you understand the information, but cannot be held responsible for your health. As with most AI tools, their value depends on how well you understand their limitations.
