ChatGPT Health: Benefit to Society or Just OpenAI?

When I first heard about the release of ChatGPT Health, I couldn’t believe my ears …

Unsure what it was, I asked ChatGPT, “What is ChatGPT Health?”

ChatGPT:

“ChatGPT Health is a dedicated space inside ChatGPT designed for U.S. users 18 and older to securely connect personal medical records, wearables, and wellness apps. It helps you understand test results, prepare for doctor visits, and track fitness data with extra privacy protections.”

At first glance, it may sound like a good idea to some. However, I see major problems with it.

To learn more, I went back to ChatGPT and typed in the following:

“Apparently, your company just made ChatGPT Health available to all US users. As you may determine, most medical professionals will not be overjoyed by this move.”

And here is the response I received from the AI:

ChatGPT:

“OpenAI announced on July 23, 2026, that Health in ChatGPT is rolling out to eligible U.S. users age 18 and older on Free, Go, Plus, and Pro plans.

It is available on web and iOS, although OpenAI says the gradual rollout may take several weeks to reach everyone.

Users can ask health questions without connecting anything, or they can connect supported medical records and Apple Health data for more personalized responses.

And yes, many healthcare professionals are likely to have a strong initial reaction.

The announcement touches several sensitive issues at once:

• Patients may treat an AI-generated explanation as medical advice, even when the product says it is not a substitute for professional care.
• ChatGPT may become the first place patients go with symptoms, laboratory results, medication questions, and treatment concerns.
• Clinicians may have to correct inaccurate conclusions, address anxiety created by AI responses, or explain why their clinical judgment differs.
• Patients may arrive with lengthy AI-generated lists of diagnoses, tests, or treatment requests.
• Connecting medical records raises understandable questions about privacy, consent, data access, security, and patient comprehension.
• Some professionals may see it as another step toward replacing or devaluing clinicians.”

While the AI acknowledged the potential concerns of healthcare providers, it noted that ChatGPT Health will change the patient-provider relationship but will not eliminate the need for skilled clinicians and presented the following arguments:

ChatGPT:

“AI can summarize a laboratory report, suggest questions for an appointment, identify trends in connected information, or explain unfamiliar terminology.

OpenAI explicitly promotes uses such as preparing for appointments, understanding test results, tracking wellness goals, and identifying questions to discuss with a healthcare professional.

But it cannot independently perform a physical examination, reliably recognize everything a patient has failed to mention, assess nonverbal cues, assume legal responsibility for care, or build the kind of therapeutic relationship that influences whether a patient accepts and follows a treatment plan.

The real question is not whether patients will use ChatGPT Health. They will. The question is whether healthcare professionals will help patients use it safely, or leave them to interpret its answers alone.” (Emphasis added by ChatGPT.)

My follow-up question to the AI was this:

“Who will reimburse healthcare professionals for their time explaining to patients how to evaluate what ChatGPT tells them? Will it be OpenAI?”

This was the response I received …

ChatGPT:

“No—there is currently no indication that OpenAI will reimburse clinicians for that time.

OpenAI positions Health in ChatGPT as a tool that helps people become more informed and prepare for conversations with healthcare professionals.

It also states that the product is intended to support, not replace, medical care and is not intended for diagnosis or treatment. But its public materials do not describe any mechanism through which OpenAI would compensate an NP, physician, pharmacist, or other clinician who must review, correct, contextualize, or explain its output.

That leaves the cost where healthcare has traditionally placed it: on the clinician, the practice, the payer, or the patient.

A clinician may sometimes be able to include this work within a legitimate E/M service when reviewing the information is medically necessary, affects assessment or treatment, and the visit otherwise meets the applicable coding and documentation requirements.

Medicare recognizes E/M services furnished by physicians and eligible non-physician practitioners, including NPs. However, simply explaining how to evaluate ChatGPT answers would not automatically create a separately reimbursable service.

Coverage would depend on the clinical circumstances, code requirements, payer policy, documentation, and possibly time or medical decision-making.

Nor does the existence of an AI-related CPT code necessarily solve the problem. The AMA notes that a CPT code does not guarantee coverage, reimbursement, or payment, and it describes the payment environment for health-related AI as still unsettled.”

The AI acknowledged the problem central to the release of the “Health CPT,” stating:

ChatGPT:

“A technology company can give millions of patients individualized health explanations, but the healthcare system may expect clinicians to absorb the cost of correcting, interpreting, and managing the consequences.

Consider what may happen in a typical practice. A patient brings in several pages of ChatGPT output. The provider must determine what information the patient entered, identify unsupported assumptions, correct inaccuracies, address the patient’s anxiety, explain why a suggested test may not be appropriate, and redirect the discussion toward the patient’s actual clinical needs. That could consume a meaningful portion of the appointment.

The AI company does not pay for that time. The insurer may not pay anything additional. Yet the provider still carries the professional responsibility and liability for the advice ultimately given.

This does not mean every patient conversation about AI should generate a special charge. Patients have always brought information from Google, family members, television, and social media.

But ChatGPT Health may change the volume, personalization, and apparent authority of that information. Reviewing a customized AI interpretation of medical records could require substantially more work than answering a quick question about something a patient reads online.

ChatGPT Health may be free or inexpensive for patients, but it is not necessarily cost-free for healthcare. Some of its hidden costs may be transferred directly to clinicians and small practices.

To what extent are AI companies creating additional uncompensated clinical work while leaving healthcare professionals responsible for resolving the uncertainty their products generate?”

Here is the irony of it all …  below the AI response, there was the following disclaimer, in small print:

“ChatGPT can make mistakes. Check important info.

ChatGPT Health leaves us with many unanswered questions:

• Who is accountable when the information is wrong, misinterpreted, or misunderstood?
• Who pays clinicians for the additional time?
• What safeguards exist for vulnerable or low-health-literacy patients?
• How will the GPT be monitored and evaluated for harm?
• Why should a technology company be able to add to healthcare-related workload without sharing responsibility for its consequences?

At times, ChatGPT is known to hallucinate and make up stuff. Compared to other AIs, it tends to provide responses geared toward pleasing or not upsetting the user.

While I am not aware of attempts to regulate or provide safeguards for such powerful AI, I think it is necessary to include them.

It appears that little or no thought is given to how this will impact an entire industry, let alone the patients who have no clinical training to evaluate the detailed information they will receive.

I regard OpenAI’s release of ChatGPT Health as highly irresponsible.

While this may be good for the company, it is not so good for the rest of society. 

What are your thoughts? Please share them with us in the comment section below.

By Johanna Hofmann, MBA, MAc., regular contributor to the NPBusiness blog.