Your data. In your hands.

1. About this policy

This Privacy Policy explains what information Mollow, Inc. (“Mollow,” “we,” “us”) collects through the Mollow Memory Mirror beta service (the “Service”), how we use and share it, and the choices you have. It applies to you when you use the Service as an invited friends-and-family beta participant.

The Service is early beta software under active development. Our data practices may change as the Service evolves; we will update this policy and, for material changes, notify you in-product. Please also read the companion Terms of Service.

2. Who to contact

Questions or privacy requests: [email protected]. Mollow is located in Providence, Rhode Island, U.S.A.

3. What the Service does with your information, in brief

Memory Mirror builds a persistent record of your conversations, decisions, and rationale across the AI tools you connect, and turns that record into your Memories, Learnings, and Habits. Doing this necessarily involves storing your content on our hosted infrastructure and sending parts of it to the AI providers that power the Service. This policy describes that flow plainly.

4. Information we collect

• Account information — the identifiers you provide to create and access your account (such as name and email), and authentication data.
• Your content — the inputs you submit and the records the Service derives from them: your conversations, uploaded files and documents, and the Memories, Learnings, and Habits built from your use.
• Connected-service data — when you connect a third-party AI tool, the content you direct the Service to exchange with it.
• Browser-extension data — if you install the Mollow Memory browser extension, the conversations it reads from the supported AI chat sites you use (see “How we collect — browser extension” below).
• Payment information — if you subscribe, billing details processed by our payment processor (we do not store full card numbers).
• Support communications — emails and messages you send us.
• Usage and device data — logs, metrics, and telemetry about how the Service is used and how it performs.

How we collect — browser extension. Some of your content reaches the Service through the Mollow Memory browser extension, which you install and sign into. With your permission, the extension runs on the AI chat sites you use (currently ChatGPT, Claude, and Gemini) and reads the conversation shown on the page so it can be saved to your Memory. After you sign in, automatic capture is on by default: as you chat, the extension periodically reads new or changed conversations on those sites and sends them to the Service. You can turn automatic capture off — globally or per site — in the extension’s settings at any time. Separately, you may use “Import past chats” to have the extension sweep your existing conversation history on a connected site and import it; this runs only when you start it, one site at a time, and you can stop it at any time. The extension reads only the supported AI chat sites listed above; it does not read other websites.

How we collect — Mollow desktop app. If you use the Mollow desktop app and sign in to a supported AI service within it (currently Claude and ChatGPT), the app reads the conversations in those signed-in sessions on your device and sends them to the Service to be saved to your Memory — the same conversations, treated the same way, as the browser extension. Capture happens on your device using your own signed-in session; Mollow does not store your username or password for those services. You can also import a chat export you download from a supported service (for example, a ChatGPT or Claude data export); imported files are processed to build your Memory and stored on our infrastructure.

5. How we use your information

We use your information to: provide and operate the Service; build and maintain your Memories, Learnings, and Habits; process the AI requests you initiate; provide support; process payments; secure the Service and prevent abuse; and improve the Service using aggregated, de-identified usage information.

6. AI processing and model training

Operating the Service requires sending your content to third-party AI providers.

• Mollow does not train its own models on your content. This is our product commitment and reflects how the Service is built.
• Your content is sent to AI providers to generate responses. Platform AI requests are routed through OpenRouter to underlying foundation-model providers; if you supply your own provider key (BYOK), requests may go directly to that provider (e.g., Anthropic). Each provider processes your content under its own terms and privacy policy.
• Upstream training. Content sent to AI providers is governed by those providers’ terms. The major foundation-model providers (for example, Anthropic and OpenAI) state that they do not train on API inputs or outputs by default. For platform text and chat requests, Mollow configures its OpenRouter organization to enable Zero Data Retention and to opt out of training — routing only to provider endpoints that do not store request data, with every train-on-data and data-sharing option turned off.
• Embeddings and document processing. To enable search, text is sent to Google Vertex AI for embeddings; uploaded documents are sent to Google Document AI for parsing.
• Voice and text-to-speech. Audio features send text to ElevenLabs. Mollow has turned off ElevenLabs’ model-training setting for its workspace; however, ElevenLabs’ terms permit it to use de-identified or aggregated data for its own product improvement and research, and zero-retention processing is not available on Mollow’s current ElevenLabs tier. Mollow therefore does not represent that text submitted for speech synthesis is excluded from de-identified training or retention by ElevenLabs.

7. Third parties and sub-processors

We share information with the service providers below solely to operate the Service.

Cloudflare is used for DNS infrastructure only; no user content is routed through it as part of the Service.

We may also disclose information to comply with law, enforce our Terms, or protect the rights, safety, and security of users, the public, or Mollow. If Mollow is involved in a merger, acquisition, or asset sale, information may be transferred as part of that transaction.

8. Sharing and the public Memory Mirror

If you choose to publish a Memory Mirror, sharing is opt-in and category-scoped, and anyone with the link can view the selected categories without logging in. Published content may be viewed, copied, downloaded, cached, or indexed by others, and revoking or expiring a link does not retract copies already made. The shareable image includes a Mollow referral link. Share links remain active until you revoke them, and published pages may be crawled and indexed by search engines. Do not publish content you are not entitled to share.

9. Storage, retention, and de-duplication

Your content is stored on Mollow’s hosted infrastructure (databases and Google Cloud Storage). To operate efficiently, we may de-duplicate byte-identical information at the storage layer. During the beta, reference material you upload may be de-duplicated into a shared storage layer; please do not upload material you are not entitled to share. We retain your information for as long as your account is active and as needed to provide the Service, comply with law, resolve disputes, and enforce our agreements.

10. Your choices and rights

• Access and correction — you can view and edit much of your content within the Service.
• Deletion — you may delete content and request deletion of your account; some information may be retained as required by law or for legitimate business purposes.
• Export / portability — account data export is not yet generally available during the beta. You may contact us to request a copy of your data, which we will provide on a reasonable-efforts basis.
• California residents — see the separate CCPA addendum for CCPA/CPRA-specific rights.

11. Security

We use commercially reasonable, best-efforts measures to protect your information. During the beta, sensitivity classifications (such as private or restricted) are labels that organize your content and signal your intent; they are not yet enforced access controls, and enforcement is being hardened during the beta. Because the Service is in beta, no method of storage or transmission is guaranteed secure, and we do not warrant against loss or corruption of content — keep your own backups of anything important.

In transit, your data is protected with TLS (HTTPS). Our infrastructure runs on Google Cloud, which encrypts data at rest at the storage layer. In addition, Mollow encrypts the secrets you entrust to it — such as the API keys you provide for your own AI-provider accounts (BYOK) — at the application layer using AES-256-GCM. To operate and provide the Service, Mollow’s systems process your message content, transcripts, and the Memories, Learnings, and Habits built from them to serve you — we do not sell your data or use it for advertising. Providing the Service also involves the AI providers and sub-processors described in Section 7, and means Mollow can access your content. The Service is not end-to-end encrypted and is not zero-knowledge.

12. Children

The Service is for adults. You must be at least 18 years old to use it, and we do not knowingly collect information from anyone under 18.

13. United States

The Service is operated from, and intended for users in, the United States. By using it you understand your information will be processed in the U.S.

14. Changes to this policy

We may update this policy as the Service evolves. We will post the updated version with a new effective date and, for material changes, notify you in-product.

15. Contact

[email protected] — Mollow, Inc., Providence, Rhode Island, U.S.A. Companion to the Beta Terms of Service.