Why Standard ChatGPT Is Not HIPAA Compliant
HIPAA compliance isn't a feature a tool either has or doesn't, it's a legal relationship. A vendor becomes HIPAA compliant for your practice's use only when they sign a Business Associate Agreement (BAA) with you, taking on legal responsibility for how they handle protected health information (PHI) on your behalf. The free version of ChatGPT, and the standard ChatGPT Plus subscription most individuals use, does not include a BAA. That means if a staff member pastes a patient's name, diagnosis, or any identifying health detail into a regular ChatGPT conversation, that's a HIPAA violation, regardless of how careful or well-intentioned the person was.
This catches a lot of practices off guard because ChatGPT feels private, like a conversation with a tool rather than a third party. Legally, it's the opposite. Every prompt sent to ChatGPT is processed on OpenAI's servers, and without a BAA in place, there's no contractual obligation covering how that data is stored, used, or protected.
When ChatGPT Can Be Used Under a BAA
OpenAI does offer BAAs, but only through specific enterprise and API-level arrangements, not the consumer product most people are used to. This typically means going through OpenAI's business offerings directly, or accessing the underlying models through a platform that has its own BAA in place with OpenAI and extends that protection to you.
Even with a BAA signed, a few things still need to be true for actual compliance. Data needs to be encrypted both in transit and at rest. Access needs to be limited to staff who actually need it, with audit logs showing who accessed what and when. There needs to be a clear answer for how long data is retained and how it's deleted. A BAA covers the legal relationship, it doesn't automatically mean the technical setup around it is secure, that still depends on how the tool is configured and used.
What a Compliant Setup Actually Requires
A practice considering ChatGPT, or any general AI tool, for anything touching patient information should be able to answer these questions clearly before using it:
Is there a signed BAA with the vendor specifically covering this use case? Is data encrypted in transit and at rest? Is there an audit trail showing who accessed what patient data and when? Is there a clear, written policy on data retention, and does the vendor actually delete data on that schedule? Are staff trained on what can and can't be entered into the tool, since even a compliant backend doesn't protect against a staff member misusing it?
If the answer to any of these is unclear, the safest assumption is that the tool isn't ready for patient data yet, no matter how useful it is for other tasks.
A Safer Path: Purpose-Built Healthcare AI
General AI tools like ChatGPT were built to be useful for everything, which is exactly why getting them compliant for one specific, high-stakes use case like healthcare takes extra work. Tools built specifically for healthcare, like the ones we implement through healthcare AI consulting, come with the BAA, encryption, and access controls built in from the start, rather than needing to be configured around a general-purpose product.
This matters most for anything patient-facing or clinically documented, things like AI scribe tools that transcribe exam room conversations, or AI patient scheduling systems that handle appointment history. These are exactly the use cases where a general consumer tool without a BAA creates real legal exposure.
What to Do If Your Team Is Already Using ChatGPT
If staff have already been using standard ChatGPT with any patient information, the first step is simply to stop and review what was entered, since that history may need to be addressed as part of a compliance review. From there, a clear written policy on acceptable AI use, covering what tools are approved, what can and can't be entered, and who to ask when unsure, closes the gap going forward. Our free AI readiness assessment is a good starting point if you're not sure where your practice currently stands.

