GOOSENEWS · PRIVATE BETA
Data Policy & Service Terms
Effective 19 August 2026 · Registration policy version 2026-08-19
DATA POLICY
How GooseNews handles data
1. Controller and scope
For the private beta, GooseNews is operated by the person or legal entity identified in the invitation that gave you access. That operator determines why and how account data is processed and is therefore the controller. Until permanent details are published, use the request form below and the invitation sender for privacy matters. This temporary description is not sufficient for an open commercial launch.
2. Data map, purpose and lawful basis
| Data used today | Why it is used | GDPR basis | Who receives it |
|---|---|---|---|
| Account and authentication Email, display name, password hash, plan/role, invite use, verification/reset token hashes, session token hashes and timestamps. | Create, authenticate, recover and administer the account. | Contract; legitimate interests in account and service security. | GooseNews hosting/database and email-delivery providers. |
| Reader settings Languages, preferred news language, home city, personal RSS URLs, folders, saved/read/hidden state, reading progress, feedback, Editorial topic weights and links to the learning editions in your private library. | Provide your reader, study shelf, entitlements, city view and account-specific Editorial ranking. | Contract; legitimate interests for optional, low-impact personalization. | GooseNews hosting/database. Not advertising networks. |
| Public source data RSS URL, source title, article title, source-provided summary and content, author, dates, original link and—where permitted and technically accessible—text extracted from the linked public page. | Fetch the public source, show original-source cards, group Editorial reporting and maintain a shared processing cache. GooseNews does not bypass logins, paywalls or access controls. | Contract and legitimate interests in providing the news reader; rights-holder interests remain applicable. | GooseNews hosting/database; selected source text may be sent to the AI provider as described below. |
| AI task data Source name, article title, selected RSS or publicly accessible page text, event context, requested language and learning level. Account email, password and payment data are not placed in prompts. | Automatically analyse approved Editorial sources into event relationships, briefs and preferred-language output; generate a Personal RSS learning edition only when you request it. | Contract for requested features; legitimate interests in preparing the Editorial service. | OpenAI API as a processor/service provider, subject to the operator's contract and configuration. |
| Security and operations Hashed rate-limit identifiers derived from IP/email, job state, allowance/cost records and audit actions. Infrastructure logs may include IP address, user agent and request metadata. | Prevent abuse, operate queues, enforce allowances, diagnose failures and evidence administrative actions. | Legitimate interests in secure and reliable operations; legal obligations where applicable. | GooseNews hosting/database and infrastructure providers; authorised operators. |
| Requests and notices Requested city, notes, reporter email, related URL, complaint details and resolution. | Review city coverage, corrections, opt-outs, copyright/illegal-content notices and data-rights requests. | Steps requested by you; legitimate interests; legal obligations and legal claims where applicable. | Authorised editors/operators and relevant advisers or authorities when necessary. |
3. What remains private and what can become shared
Personal subscriptions, folders, reading state, preference signals, language choices, city requests and the list of editions on your Learning Library shelf are account-private by default. A personal feed is not placed in the Editorial candidate queue unless you select Suggest to editors. If you do, editors can review the public feed; unnecessary submitter identity should not be shown in the editorial view.
Public RSS feed records, public article metadata, de-identified event clusters and cached AI editions are shared service data and may remain after an account is deleted. They are not intended to contain the former user's account identity.
4. AI provider handling
GooseNews sends text tasks through the OpenAI Responses API with application storage disabled (store=false). OpenAI states that API data is not used to train its models by default unless the customer opts in. Under standard controls, prompts, responses and related metadata may still appear in abuse-monitoring logs for up to 30 days, subject to OpenAI's legal and safety exceptions. GooseNews does not claim Zero Data Retention or EU-only processing unless the operator separately enables and verifies those controls.
Before public launch, the operator must confirm that training-data sharing remains disabled, execute the applicable data-processing terms, document subprocessors and Chapter V transfer safeguards, and decide whether to obtain OpenAI's European data residency and Modified Abuse Monitoring or Zero Data Retention controls.
5. Retention and deletion
- Login sessions expire after at most 30 days; expired sessions are purged by the worker.
- Expired email-verification and password-reset records are removed after a seven-day operational buffer; rate-limit buckets after roughly two days; completed background jobs after 30 days.
- Account-bound subscriptions, settings, reading state, Learning Library links, language slots, city requests and allowance records are deleted when the account is permanently deleted.
- Shared public-source records and de-identified cached outputs may remain while the source is active or the cache remains useful.
- Audit and complaint records are kept only as long as reasonably necessary for security, accountability, notice handling and the establishment, exercise or defence of legal claims. The operator must document and periodically review the concrete schedule before public launch.
6. Personalization and automated processing
GooseNews can combine your explicit category weights with read, hide and “more/less like this” signals to rank Editorial events. It does not make decisions producing legal or similarly significant effects. You can disable personalization and clear learned signals in Account Settings without deleting saved articles.
7. Cookies and device access
The current service uses one first-party, secure, HTTP-only authentication cookie that is necessary to keep you signed in. It does not install advertising, cross-site tracking, analytics or geolocation cookies, and browser permissions for camera, microphone and geolocation are disabled. If non-essential tracking is added later, it must remain off until a separate valid choice is obtained.
8. Your rights
Subject to the GDPR and applicable limits, you may request access, correction, deletion, restriction, portability or object to processing based on legitimate interests. You may also complain to Luxembourg's CNPD. Account Settings provides direct profile correction, personalization controls and permanent account deletion. For other requests, choose Other in the form below, begin the details with “Data rights request”, and provide enough information for identity verification. Requests are handled without undue delay and normally within one month.
9. Security, children and changes
Current safeguards include Argon2 password hashing, hashed login/reset tokens, secure production cookies, same-origin checks, tenant-scoped database access, restricted RSS fetching and encrypted transport. No internet service can promise absolute security. The beta is for users aged 16 or older and does not knowingly solicit special-category personal data. Material policy changes require a new version and, where legally required, a new notice or choice.
TERMS OF SERVICE
Rules for using the beta
Account and access
You must be at least 16, provide accurate registration information, protect your password and use only your own invitation. Private-beta access may be limited or withdrawn for security, abuse, legal or operational reasons.
The service
GooseNews follows public RSS/Atom sources, groups approved Editorial reporting by event, and provides allowance-based language-learning editions. Availability, source coverage, model output and beta features can change. Generation is not unlimited; a cached edition may be reused without another allowance charge.
Your responsibilities
Do not submit credentials, cookies, tokens, private-network addresses, paywall bypasses, unlawful material, malicious feeds or sources you are not entitled to access. Do not probe, overload or circumvent security, quotas or plan restrictions.
AI and news accuracy
AI-prepared titles, briefs, translations and learning editions can contain errors or omit context. They are labelled as AI-assisted, are not professional advice, and do not replace the publisher's report. Keep the original link and verify consequential claims with the original source.
Paid service
No payment-card collection or paid checkout is implemented in this beta. Before charging consumers, GooseNews must publish the trader identity, total price and taxes, renewal/cancellation terms, performance details, complaint route and any applicable withdrawal information, and must use a payment processor with a separate notice.
Deletion and mandatory rights
You may permanently delete your account in Account Settings. Nothing in these terms excludes non-waivable EU or Luxembourg consumer, data-protection or other statutory rights. Governing-law, trader and dispute-resolution clauses must be completed after the operator entity is fixed and reviewed before a paid launch.
CONTENT & COPYRIGHT
Original sources come first
By default, GooseNews shows the source, title, publication date, original link and a short source or AI-prepared event summary. For approved Editorial processing—and for a Personal RSS learning edition you explicitly request—it may temporarily fetch and store text from a publicly accessible linked page so that clustering and synthesis use the underlying reporting. It does not intentionally cache source images, display stored full source text, bypass access controls or ask the model to reproduce an article. Generated reading dossiers must synthesize and independently paraphrase multiple reports, link to the originals and avoid serving as a substitute for them. Editorial sources require a recorded rights/processing decision; publishers may request correction, catalog opt-out or removal.
RSS availability does not by itself grant every reuse right. Hyperlinks, facts and very short extracts receive different treatment under EU law, while longer excerpts, translations and commercial aggregation can require permission. The operator must continue source-by-source rights review and prompt notice handling.
REQUESTS & COMPLAINTS
Submit a request
Use this form for copyright, illegal-content, correction and publisher opt-out notices. For a privacy request, choose Other and begin the details with “Data rights request”. Do not include passwords, tokens or unnecessary sensitive data.