Skip to main content

HIPAA & Privacy

How we protect your health information

USDOCAI is built around the HIPAA Privacy, Security and Breach Notification Rules. This page explains our safeguards and includes our Notice of Privacy Practices.

Effective date: September 1, 2025

Our safeguards

Three layers of protection

Administrative

  • A designated Privacy Officer and Security Officer
  • Regular HIPAA risk assessments and workforce training
  • Business Associate Agreements with every vendor that handles health information
  • Provider licenses verified before a clinician can see patients
  • A documented breach response plan

Physical

  • Data stored in HIPAA-eligible cloud data centers with controlled physical access
  • No health information kept on your phone; optional biometric login is encrypted in the device keystore
  • A privacy screen hides your information in the phone’s app switcher

Technical

  • Encryption in transit (TLS) and at rest
  • Least-privilege access: only you and your treating providers can open your record
  • Automatic sign-out after 15 minutes of inactivity
  • Audit logs of every view and change to health information
  • Signed clinical notes and consent records can’t be altered

AI and your data. The USDOCAI Assistant’s symptom screening runs on your phone. If you continue to a visit or choose to share it in a chat, a short summary (your symptom, how long, and your temperature if you measured it) goes to your doctor. If you use voice input, the recording is turned into text in our own secure cloud and deleted right away. Photos, files and voice messages you send can be opened only by you and your care team. To find a pharmacy, we send only your ZIP code to the national NPI Registry. The AI scribe only records and transcribes a visit with your consent. Speech recognition runs in our own secure cloud, not at a third-party transcription service, and the recording is deleted as soon as it’s transcribed. Your doctor reviews and signs every note, and your recordings and notes are never used to train AI models. When the scribe checks a suggestion against public references (FDA drug labeling and the National Library of Medicine), it sends only the medication, test or diagnosis name, never anything that identifies you.