Privacy Policy
Who is responsible
Last updated: October 1, 2026. Effective when published.
Brian Bell operates reading-lite at readinglite.ai and readinglite.app and is responsible for the personal information described here. Contact operator@readinglite.app with privacy questions or requests.
Information we collect
- Account information: your email address, account settings, authentication records, sessions, API-token records, and device or browser labels used to help you recognize sessions.
- Provider key: if you connect your own OpenAI API key, we store it encrypted, with its last four characters, whether OpenAI accepts it, and when it was checked. We also record estimated charges against your daily spending limit.
- Saved material: submitted URLs and documents, downloaded source content, extracted text, summaries, tags, search representations, and connections between readings.
- Operational information: processing and error records, usage-limit records, email-delivery records, and information you send us when requesting support. Our infrastructure providers also process connection information, such as IP addresses, to deliver and protect the service.
How we use information
We use this information to authenticate you, save and search your readings, generate summaries, send requested messages, troubleshoot failures, enforce usage limits, prevent abuse, and respond to support and legal requests. We do not sell your personal information or use your saved readings for targeted advertising.
Where applicable law requires a legal basis, we process information to provide the service under our agreement with you, for legitimate interests in securing and maintaining the service, to meet legal obligations, or with consent where required. You can withdraw consent for optional processing without affecting earlier lawful processing.
Who receives information
- Cloudflare hosts the website, application, account database, saved content, search index, and background processing. Its Turnstile service may help protect sign-in from abuse.
- OpenAI receives text from processed readings to create search representations. The configured summarization provider, OpenAI or Anthropic, receives reading text and relevant material from earlier readings in your archive to generate summaries. The service does not silently switch summarization providers after a failure.
- When your account uses your own OpenAI key, those requests are made with your key, OpenAI handles them under the terms and policies of your OpenAI account, and OpenAI bills you for them. Checking a key when you connect it asks OpenAI which models the key can use, which is not charged.
- Resend receives email addresses and message content to deliver sign-in links, security messages, and reading notifications. Reading notifications can contain information from your archive.
- When you submit a URL, the source website receives a request for that page. Some imports also contact platform APIs, including YouTube or X, with the URL or identifiers needed to retrieve the material.
AI processing takes place outside your device. Provider retention and any use of submitted data are governed by the applicable API agreements and provider policies; this service does not promise zero retention by those providers. Do not submit secrets or other material you are not comfortable sending to these providers.
We may access account information and saved material when needed to provide support, investigate abuse or failures, or meet legal obligations. We may disclose information when legally required or reasonably necessary to protect rights and safety. Copyright complaints and responses may be shared with the other party so the claim can be addressed.
Security and location
Accounts are separated using access controls, and your readings are not made available as public pages or shared with other users through the service. Your archive is not end-to-end encrypted: the application and its providers must process readable content to deliver its features. No online service can guarantee complete security.
A provider key you connect is encrypted with AES-256-GCM under a server secret kept apart from sign-in secrets, and bound to your account so it cannot be used for another. It is decrypted only in memory while a request to OpenAI is made. We never show it again in full: the service, its emails and exports, and our logs show at most its last four characters. Sign-in sessions can manage the key; API tokens cannot read or change it.
Our providers may process information in the United States and other countries, which may have different privacy protections from your country. Where applicable law requires safeguards for international transfers, those requirements apply to our processing. Contact us for information about the arrangements relevant to your account.
How long information stays
Saved readings do not expire automatically. We generally retain them while your account remains open unless you delete them or we remove them under the Terms. Failed imports and processing records may also remain with your archive. Email-digest delivery records remain after a message is sent, skipped, or fails, and are removed during account teardown.
A limited account-deletion audit record, including the account identifier, deletion dates, and cleanup status, remains after account teardown so we can track deletion attempts and investigate incomplete cleanup. It does not contain your saved article text or email address. That record currently has no automatic expiry.
Deleting readings or your account
You can request help deleting a reading or your account at operator@readinglite.app. Deletion is also available through the authenticated API. The web app does not yet have deletion buttons. We may need to verify your identity before acting.
Removing your provider key on the Provider page deletes the stored key at once. Readings already saved stay readable and searchable without it.
Starting account deletion immediately disables sign-in, deletes any stored provider key, and revokes sessions and API tokens. Cleanup of saved content and search data normally starts after about seven days, unless we start it sooner. This delay is not a promise that the account can be recovered.
Cleanup runs in the background. Search data can take additional time to disappear, and incomplete cleanup can leave stored copies or search data requiring operator attention. We cannot currently give a guaranteed completion date or promise that every residual copy has been removed. Contact us to ask about the status of your request.
Deleting an individual reading removes it from your library and initiates removal of its stored content and search data. Deletion may need to wait while that reading is being processed. Earlier summaries or connections in your other readings can still contain references to it; those are not automatically rewritten. Deleting from this service does not remove the original source website.
Your choices and requests
You can view your readings and retrieve individual saved documents through the website or API. There is no built-in bulk account export. Save anything you need before requesting account deletion, because access ends immediately when deletion starts.
Depending on where you live, you may have rights to access, correct, delete, or receive a portable copy of personal information, restrict or object to processing, withdraw consent, or complain to a privacy regulator. Contact operator@readinglite.app to exercise applicable rights, including when the app has no corresponding control. We will verify requests as appropriate and respond within the period required by applicable law. A missing product feature does not limit your statutory rights.
Children
The service is intended for adults, not children under 18. If you believe a child has provided personal information, contact us so we can investigate and address it.
Policy changes
We will update this policy and its date when our practices change. We will provide notice through the service or email before material changes take effect and obtain consent where required. See Terms for the agreement governing use and Copyright for copyright complaints.