# HUVA Privacy Policy

Version: 2026-11-05.1 · Effective November 5, 2026

Huver, Inc., a Delaware corporation ("HUVER," "we," "us"), operates HUVA Work, HUVA Scribe and Scribe Lite, HUVA Chat, and the HUVA website. This policy describes personal information handled through those services in the United States. The product sections below explain important differences. Our address is 1111B South Governors Avenue, STE 34419, Dover, DE 19904, United States. Send privacy questions, requests, and appeals to **issong@huver.ai**, with the subject "HUVA Privacy."

## 1. Our role and healthcare information

For website visitors, account administration, billing, and professional research in Chat, HUVER handles information for the purposes described in this policy. For patient information processed on behalf of a healthcare organization, we act under its lawful instructions, the applicable service agreement, applicable healthcare law, and a Business Associate Agreement (BAA) where required. The organization's Notice of Privacy Practices explains rights in its medical records. This policy does not replace that notice, a BAA, patient recording permission, or any required authorization.

Patients and their representatives should contact their healthcare organization about medical records. We assist that organization with information we hold for it. Work and Scribe may process children's patient information under the organization's authority; our professional accounts and Chat are intended for adults, not for children to use independently.

## 2. Information we collect and its sources

We receive information from you, your representatives, healthcare organizations, shared HUVA account services, connected systems, and service providers. We also receive technical information when you use a service. Collection depends on the functions used.

- **Accounts and organizations:** names, business and contact details, profile images, sign-in identifiers, professional qualifications or verification materials, organization memberships, roles, clinic settings, integration credentials, and messaging-registration business and representative information.
- **Agreements and preferences:** the versions acknowledged or accepted, the account identifier, signer name, organization legal name, stated signing authority or user capacity, server acceptance time, recording disclosures, and messaging consent and opt-out records. An individual account acknowledgment and an organization representative's BAA acceptance have different effects, as explained in the acceptance screen and BAA.
- **Work:** patient identifiers and contact details, dates of birth, sex, insurance and subscriber information, appointment requests and bookings, intake answers, signatures, consent records, documents and images, clinical context and notes, calls and recordings where enabled, transcripts, summaries, messages, delivery records, and workflow actions. Insurance eligibility requests may include the requester's IP address.
- **Scribe:** relevant information from the selected EHR, including patient and encounter identifiers, history, medications, allergies, supported prescription context, clinician notes, encounter audio, transcripts, draft notes, coding suggestions, edits, and EHR entry and confirmation results. Device pairing, recovery, and recording activity are also processed.
- **Chat:** prompts, responses, conversation history, uploaded files and images, generated content and files, research queries and references, optional location queries, shared conversations and their links, feedback, and processing records. **Do not submit identifiable patient information to Chat.**
- **Billing and activity:** billing contacts, subscription and payment status, transaction references, credit and usage records, device and browser information, connection data, timestamps, diagnostics, support communications, cookies, and browser storage. Card details are collected on payment-provider pages; HUVA receives the payment references and status needed for billing. Clinical audio and text must not be included in payment metadata.

## 3. Why we use information

We use information to authenticate users, manage roles and organizations, record agreements, provide purchased and enabled functions, maintain account history, process selected content, communicate as instructed, manage subscriptions and support, investigate failures, protect services, respond to privacy requests, and meet legal and contractual obligations.

Work uses relevant information for intake, clinical workflows, supported patient registration and appointment actions, calls and messaging, insurance eligibility where enabled, and receptionist setup and testing. Scribe uses relevant encounter context and audio to produce draft documentation, support review and requested EHR entry, recover interrupted work, and administer usage. Chat uses non-patient-identifying content for professional information, research, file processing, optional sharing, and supported maps.

HUVER does not provide medical care or make insurance coverage determinations. AI output can be incorrect. Authorized healthcare personnel must review clinical and coding output before using it in patient care, a medical record, or a claim. Routine administrative actions may operate within the healthcare organization's approved rules and exception paths.

## 4. Recipients and service providers

The healthcare organization, its authorized users, and recipients it lawfully designates receive information for their workflows. Depending on the enabled function, these recipients may include patients, representatives, selected EHRs, healthcare providers, payers, and insurers.

Service providers receive information needed for their functions. Current examples include Google Cloud and Google AI services for hosting and AI processing; Clerk for sign-in; Stripe for payments; Twilio for supported communications and messaging registration; Stedi for insurance eligibility; and Resend for email delivery. This list describes functions, not a representation that every provider or product is authorized to receive PHI. A PHI-processing path must be covered by the applicable agreements, provider service scope, and configuration before use. General account, payment, marketing, and support services must not receive clinical content merely because they support the platform.

Scribe can send relevant diagnosis or procedure search terms to reference services such as the National Library of Medicine's Clinical Tables. Chat research can send derived queries to Google Search and medical-reference sources, including the National Library of Medicine, ClinicalTrials.gov, and FDA resources. Optional map functions may use Google Maps or OpenStreetMap's Nominatim service. Independent services you connect have their own notices and agreements.

If you enable a Chat share link, anyone with the link may access the shared content. Revoking a link does not erase copies another person has saved. We may disclose information when lawfully instructed, required by applicable law or valid legal process, or permitted for a corporate transaction subject to applicable confidentiality, healthcare, and extension-data restrictions.

## 5. AI, training, and processing locations

Requested AI functions may transmit audio, selected files, prompts, and relevant context to an approved provider. HUVER does not use PHI or PHI-derived data to train, fine-tune, evaluate, or improve general-purpose, shared, or cross-customer AI models. Customer-dedicated training or de-identification of PHI requires the authorization and conditions in the applicable BAA. This policy does not authorize such additional uses.

Chat currently uses Google's paid Gemini Developer API for non-PHI research. Under Google's paid-service terms, submitted prompts, files, and responses are not used to improve Google's products. Provider conversation state and safety, search-grounding, and service records can have separate retention periods; deleting a HUVA conversation does not automatically delete all provider records. See [Google's Gemini API terms](https://ai.google.dev/gemini-api/terms) and [Interactions data retention](https://ai.google.dev/gemini-api/docs/interactions#data-storage-retention). These Chat services are not an approved path for identifiable patient information or clinical decision-making.

Primary Work databases and file storage are in the United States. Some AI processing uses global infrastructure. Required data-location and access restrictions continue to apply. Routine access to PHI from outside the United States must follow the applicable BAA and written implementation arrangements, including the permitted countries, entities, roles, purposes, and safeguards. A Google Cloud BAA alone does not authorize every Google product, other provider, or overseas support arrangement.

## 6. Recording, extensions, and patient communications

Healthcare organizations and users must provide required notices and obtain applicable permission before recording, transcribing, or contacting a patient. Acceptance of account terms or a browser microphone prompt does not replace patient permission. Setup, preview, and test conversations can also be processed by AI and retained in operational records; use synthetic information in environments not approved for PHI.

Scribe recording begins when the user starts recording and permits microphone access. Depending on the product, audio is uploaded in segments or after recording stops. Closing the Lite side panel alone does not stop recording; use the recording control. The extension reads supported EHR pages for the disclosed encounter workflow and requested EHR actions. Some deployments request broad website access to support organization-specific EHR addresses. That permission is not permission for unrelated browsing surveillance.

The extension stores authentication, recovery, and disclosure state. Raw audio and clinical text are held in browser memory rather than Chrome local or synchronized storage. An optionally paired phone can access the selected encounter for supported recording, review, editing, or entry requests. The phone stores pairing and recovery credentials in tab session storage; disconnecting it does not erase information already processed on the server.

HUVER's use of information handled by its extension complies with the [Chrome Web Store User Data Policy, including Limited Use](https://developer.chrome.com/docs/webstore/program-policies/limited-use). We do not sell extension data or use it for advertising, data brokerage, or creditworthiness. Human access to extension content is limited to explicitly authorized support, necessary security work, or legal requirements. Any permitted business-transfer use remains subject to the policy's applicable consent requirements.

Reply STOP to opt out of supported text messages or HELP for assistance. The healthcare organization controls permitted channels and patient communication instructions. An opt-out does not delete an account or medical record.

## 7. Retention, deletion, and records

We retain information for its service purpose and applicable legal or contractual obligations. The following periods describe particular stores; they are not a single deletion deadline for all copies.

- **Work records:** organization-owned patient, appointment, intake, communication, transcript, and note records do not currently have a general automatic expiration. They are retained for the organization's workflows and applicable obligations, and return or deletion is coordinated under its lawful instructions and agreement. Work processing caches can expire separately, generally 24 hours after the relevant completed processing record. Temporary audio files may remain until their processing resources are discarded or replaced; normal completion is not a guarantee of immediate deletion.
- **Scribe Lite:** temporary encounter workspace state generally expires eight hours after creation. Encrypted transcript and generation results used for retry recovery expire separately, generally eight hours after the relevant processing record is updated. Raw audio is used for transcription and transient request processing and is not stored as part of the Lite encounter database or Redis record. Browser memory and provider processing are separate.
- **Chat:** conversations, uploads, generated files, feedback, and shares are retained for account history and requested features. Deleting a conversation removes its active HUVA conversation record but does not automatically remove separately stored uploads or provider history. Account deletion has the broader effect described below.
- **Accounts and operations:** billing, usage, agreement, security, and record-integrity information may be retained independently of an encounter or conversation for administration, accounting, disputes, fraud prevention, and legal obligations. Agreement evidence may remain after account closure. Logs, provider records, protected backups, and stored versions follow their applicable purpose and retention cycles. Legal holds or other lawful requirements may delay deletion; retained information remains protected.

The healthcare organization's designated EHR remains its authoritative medical-record system unless a written agreement identifies another system. Temporary Scribe storage is not a long-term medical archive. A Confirm, Write Note, Lock Note, logout, or unlink action does not by itself erase all copies. Deleting a HUVA copy does not delete an independent healthcare provider's record. At termination, applicable BAA export, return, destruction, and legally required retention provisions govern.

## 8. Account closure and choices

Delete account in settings closes the shared HUVA account and ends its access to Work and Chat. The process removes or replaces personal profile information and removes personal Chat conversations, shares, feedback, uploads, and professional-verification materials from the active service, subject to applicable preservation requirements. There is no account-recovery grace period.

Account closure does not close a healthcare organization, cancel its subscription, delete its patient or business records, or immediately erase separately retained Scribe information. Limited closure identifiers and agreement evidence can remain for security, record integrity, or legal obligations. Uninstalling the extension, stopping recording, revoking browser permissions, signing out, or unpairing a phone prevents or limits future use but does not delete information already sent to HUVA or an EHR.

Cookies and browser storage support sign-in, session continuity, security, and preferences. Blocking necessary storage can disrupt these functions. We do not track activity across unrelated websites for targeted advertising. Browser Do Not Track signals do not change essential service processing. Our no-sale and no-cross-context-advertising-sharing practices apply whether or not a browser sends Global Privacy Control. Any new optional tracking will require the notices and choices applicable to it.

## 9. Privacy rights and California notice

Contact issong@huver.ai to request access, correction, deletion, a portable copy, an accessible policy copy, or an appeal. Identify the product and organization, if applicable; do not include medical records in the initial email. We verify identity and representative authority only as appropriate. We explain applicable exceptions and refusal reasons and do not unlawfully discriminate or retaliate for exercising privacy rights. Requests are free when required by law. Patients should direct medical-record requests to their healthcare organization; we assist it under the applicable agreement. You may also complain to the relevant regulator or, for HIPAA matters, the HHS Office for Civil Rights.

Where the California Consumer Privacy Act applies, nonexempt account, business-contact, billing, professional, and Chat information is covered. Qualifying PHI and medical records are subject to the applicable statutory exemptions; an exemption does not remove healthcare confidentiality duties. The categories, sources, purposes, recipients, and retention criteria are described above. These categories include identifiers and customer records, commercial information, internet or network activity, electronic/audio/visual content, professional information and nonclinical inferences, and sensitive account credentials or access information.

During the preceding twelve months, relevant business-purpose disclosures included identifiers to hosting and sign-in providers; transaction information to payment providers; permitted content to hosting and AI providers; communication details to messaging or email providers; eligibility information to eligibility providers; search or location queries to reference or optional map providers; and workflow information to the healthcare organization, designated EHR, or other authorized recipient. The categories apply only to the extent not legally exempt.

We do not sell personal information or share it for cross-context behavioral advertising, including information about consumers under sixteen, and did not do so in the preceding twelve months. Sensitive information is used for requested services, security, and legal compliance. Where applicable, you may request to know, access, correct, delete, or obtain a portable copy, opt out of sale or advertising sharing, and limit sensitive-information use. Authorized agents may submit requests; sale or sharing opt-outs do not require identity verification. Our current practices do not require you to opt out of sale or advertising sharing.

We acknowledge covered access, correction, and deletion requests within ten business days and respond within forty-five calendar days, or within ninety days when a permitted extension and notice apply. Other jurisdictions' applicable rights and response periods also apply. We will provide required notices and choices before materially changing these practices and review this policy at least annually.

## 10. Security and updates

We use access controls, encryption, and other safeguards appropriate to the service and applicable obligations. No system is guaranteed risk-free. Security incident notices are governed by applicable law and agreements. The public Security Overview explains the scope of HUVA's compliance statement; it is not government certification or proof that a particular organization has completed its implementation requirements.

We update this policy when practices or law change and provide additional notice or obtain consent when required. The version shown above identifies this publication. For privacy or security concerns, contact issong@huver.ai; request a secure channel before sending sensitive details.
