0000 · 0000

Dappasol / Guides

Updated June 2026

HIPAA for AI-Built Healthcare Apps

HIPAA applies if your app creates, stores, or transmits protected health information, no matter what tool built it. AI-built healthcare apps routinely fail on encryption, access controls, audit logging, and signed Business Associate Agreements with every vendor, including any LLM or hosting provider that touches PHI.

AI coding tools like Lovable, Bolt, Cursor, Replit and v0 will ship you a healthcare app that demos beautifully. HIPAA does not care how the code was written. If your app handles protected health information (PHI), the rules that always applied still apply, and in an AI-built app the gaps land exactly where HIPAA is strictest: access control, encryption, audit trails, and vendor agreements. This guide covers what HIPAA actually requires in plain terms and where AI-built apps fall down. It is general information, not legal advice. Confirm your specific obligations with a qualified healthcare attorney or compliance advisor.

Does HIPAA apply to my app? (the PHI test)

Two things decide it: who you are and what data you handle. The U.S. Department of Health and Human Services (HHS) enforces the rules. They apply to covered entities (health plans, healthcare providers, and clearinghouses) and to their business associates, which is most software vendors that handle PHI on a covered entity’s behalf.

PHI is individually identifiable health information. Here is the working test: does your app create, store, receive, or transmit health information that can be tied to a specific person? That covers the obvious stuff like diagnoses, lab results, and treatment notes. It also covers identifiers attached to health context, like names, dates, contact details, and account identifiers tied to care.

The HIPAA essentials

HIPAA’s requirements come mainly from the Privacy Rule and the Security Rule. The Security Rule, which governs electronic PHI, splits into administrative, physical, and technical safeguards. For an AI-built app, the items below are where the engineering work actually lives.

Where AI-built apps fail HIPAA

The failures are predictable, because AI tools optimize for an app that runs, not one that is compliant. These are the gaps we see most often in AI-built healthcare apps.

GapWhat it looks likeHIPAA area at risk
Row-Level Security offDatabase tables holding PHI are readable across users, or access is checked only in the clientAccess controls
PHI sent to an LLM with no BAAPatient data is passed to an AI API that has no signed Business Associate AgreementBusiness Associate Agreements
Exposed keysAPI keys or database credentials shipped to the browser or committed to git historyAccess controls, encryption
No audit trailThe app does not record who viewed or changed PHI, so access cannot be reviewedAudit logging
PHI in logs or errorsPatient data written to plaintext logs or returned in verbose error messagesEncryption, minimum necessary

The LLM gap deserves its own callout. It is one line of code to wire patient data straight into a general-purpose AI API, and people do it without thinking. But unless that provider has signed a BAA and offers a HIPAA-eligible configuration, sending PHI there is a compliance problem on its own. Same logic for your hosting and database providers.

A HIPAA readiness checklist for AI-built apps

Use this to find the obvious gaps. It is not a substitute for a formal risk analysis, which the Security Rule requires and which a qualified advisor should help you complete.

  1. Confirm scope. Figure out whether your app handles PHI and whether you are a covered entity or business associate. If you are unsure, get a legal read before launch.
  2. Enable access controls. Turn on Row-Level Security (or equivalent) on every table holding PHI, enforce auth on the server, and apply minimum-necessary access by role.
  3. Encrypt everything. TLS for all traffic and strong encryption at rest for PHI, backups included.
  4. Add audit logging. Log access to and changes of PHI, with user identity and timestamp, and keep the logs.
  5. Sign BAAs. Get a signed BAA with every vendor that touches PHI, including hosting, database, and any AI or LLM provider. If a vendor will not sign one, get it out of the PHI path.
  6. Lock down secrets. Move every key server-side, strip anything sensitive from client code and git history, and rotate exposed credentials.
  7. Keep PHI out of logs and errors. Make sure patient data never lands in plaintext logs or error responses.
  8. Prepare for breaches. Have a documented breach-response plan that meets the notification timeframes before you go live.

For the wider pre-launch picture, see our guide on whether your AI app is production ready. If you are weighing formal certifications, our overview of SOC 2 for AI-built apps covers neighboring ground, and the specific risk of exposed API keys in AI-built apps is one of the most common things we find in a healthcare build.

Want us to run this audit for you?

We do a free 15-minute build audit: you show us your AI-built app, we tell you the specific security and production gaps and what it takes to fix them. No obligation.

Book your free build audit

FAQ

Does HIPAA apply to an app built with AI tools?

Yes, if the app creates, stores, or transmits protected health information. HIPAA applies based on the data and your role (covered entity or business associate), not the tool that wrote the code. An AI-built app meets the same requirements as any other.

Can I send patient data to an AI or LLM API?

Only if that provider has signed a Business Associate Agreement and offers a HIPAA-eligible configuration. Sending PHI to a general-purpose AI API with no BAA in place is a compliance problem. Confirm the vendor's terms before any patient data is involved.

What are the most common HIPAA gaps in AI-built apps?

Access controls left open (like Row-Level Security disabled), PHI sent to an LLM with no BAA, exposed API keys, missing audit logging, and patient data written to plaintext logs or error messages. They cluster in the areas HIPAA is strictest about.

Is this guide legal advice?

No. This is general information to help you find common gaps. HIPAA obligations are fact-specific, and the Security Rule requires a formal risk analysis. Confirm your specific requirements with a qualified healthcare attorney or compliance advisor.

Book a free 15-min build audit →