Privacy and Submission Notice

Version: 2026-09-02 · Consent identifiers recorded with each form: suggestion-terms-2, suggestion-privacy-2, submission-terms-2

This notice explains what News A/B collects, why, where it is kept, for how long, who can see it, and what you can ask us to do. It covers reading the site, suggesting a topic, and submitting a report. The English text is the authoritative version; translations are provided for convenience and the English text prevails where they differ.

1. Who operates this site

News A/B is operated by an individual, not a company. Requests about your data go to the contact address shown in the "About this site" panel on every page. Where the phrase "we" appears below, it means that operator.

2. Reading the site

• No account, no cookies, no analytics. Public pages set no cookies and load no analytics or advertising scripts.
• Browser preferences. Your theme and language choices are stored only in your browser's local storage under the key newsab.prefs. Nothing in that storage is sent to us.
• Hosting. The site is served through Cloudflare. Like every host, Cloudflare receives your IP address and request headers to deliver pages and defend against abuse, and handles them under Cloudflare's privacy policy. We do not receive per-visitor logs from Cloudflare and we do not build visitor profiles.

3. Suggesting a topic

3.1 What we collect

When you send a topic suggestion, we receive only the fields on the form: the topic in one sentence, the two media groups, an optional time and language range, up to five reference links, whether you want to be credited publicly, an optional name, and an optional email address. We also record the version identifiers of this notice you agreed to and a hash of the form used to detect duplicate sends.

We do not record your IP address, request headers, or the human-verification token. Reference links are stored as text only; we do not fetch them when you submit. Submitting a suggestion invokes no AI model.

3.2 Why we use it

• Topic, media groups, range, and links: to decide whether to produce a comparison.
• Name and "credit me publicly": to credit you on the resulting topic page if you asked for that.
• Email: only to reach you if we need to ask about your suggestion. We do not send newsletters or share it.

3.3 Where it goes and who sees it

1. Your suggestion is written once into a private Cloudflare D1 database. It is never edited in place; only status events are appended.
2. We periodically pull unprocessed suggestions to a private working copy on our own machine. Names and email addresses are excluded from the review panel and from the default working context of any AI assistant we use; scripts show them only to a human who explicitly asks.
3. Email addresses and names never enter Git or any public page unless you ticked "credit me publicly", in which case the name you entered appears in the contributor line of the page record for the resulting topic.

3.4 How long we keep it

• The cloud record is deleted automatically no later than 90 days after receipt, and earlier when we decide on it or for privacy or abuse handling.
• The private working copy is kept for as long as it is needed to decide on the suggestion and, if a topic is produced from it, as provenance for that topic. A suggestion that is declined or marked as spam is not used for anything else.
• After deletion we keep an event record that contains no text you typed, only the suggestion's opaque identifier and the event.

3.5 What a suggestion is not

A suggestion is not a commission. We do not guarantee that a suggestion will be adopted, processed, or answered, and the receipt you see does not offer a status query.

4. Submitting a report

Report submission is by invitation. You produce a report with the open-source toolkit, pack it into a submission archive, and upload the archive through this site.

4.1 What we collect

• The archive. It contains the report artifacts, the pinned provenance of every pipeline stage, and the private source snapshots the toolkit collected. It may therefore contain verbatim third-party news text.
• The archive's self-description. Submission identifier, operation (create or withdraw), toolkit version, topic identifier, requested languages, your sponsor credit choice (anonymous, or a display name you chose), your source-responsibility statement, the terms version you accepted, and the archive's size and hash.
• Form fields. The single-use invitation code, an optional email address, and, for a withdrawal, the withdrawal key.
• Withdrawal key. When a report is first received, the site issues you a withdrawal key. We store only a hash of it; we cannot recover or reissue a lost key.

We do not record your IP address, request headers, or the human-verification token.

4.2 Why we use it

• The archive and its self-description: to verify, audit, re-render, and, if approved, publish the report, and to keep an audit trail of what was published and why.
• Sponsor credit: to credit you on the published page exactly as you chose. An anonymous sponsor is shown as anonymous.
• Email: only to reach you about this submission, for example when it is rejected. We do not send newsletters or share it.
• Invitation code and withdrawal key hash: to accept only invited uploads and to let the holder of the key, and no one else, withdraw the report.

4.3 Where it goes and who sees it

1. The archive is uploaded directly from your browser to a private Cloudflare R2 bucket. The bucket is not publicly readable. The intake service never opens or executes the archive.
2. We pull the archive to a private store on our own machine. Once the pull is verified, the cloud copy is deleted immediately. An archive that is never pulled is deleted from the cloud no later than 14 days after upload.
3. On our machine the archive is treated as untrusted data: it is inspected without executing anything in it, and the report is rebuilt from the artifacts using our own copy of the toolkit. Contact addresses, credentials, and signed URLs are excluded from the review panel and from the default working context of any AI assistant we use.
4. A published report shows its topic page, the sponsor credit you chose, and which AI models performed which pipeline stage. It never shows your email, the invitation code, the withdrawal key, or the private source snapshots.

4.4 How long we keep it

• Cloud copies of the archive: until the verified pull, or at most 14 days.
• Submission record in the intake database (self-description, email if given, key hash): for as long as the submission or the publication it led to exists, so that a withdrawal can be verified.
• Private store on our machine: submission archives, audit records, and publication records are kept permanently as the provenance of what this site published. Personal fields in them are handled under section 6.

4.5 Licence to your contribution

By uploading an archive you confirm that you are entitled to submit it and you grant us a non-exclusive, worldwide, royalty-free licence to reproduce, verify, re-render, translate, publish, distribute, and archive your own contribution: the report, its structure, annotations, analysis, and any text you wrote. The licence to display the report publicly lasts until it is withdrawn under section 4.7 or removed under section 4.8; the licence to keep it in our private audit records is permanent.

You cannot grant us rights you do not hold, and we do not ask for them:

• Third-party source material inside the archive, such as news articles and their snapshots, is not licensed to us by this notice. We use it only to verify quotations and provenance, we do not republish full articles, and published pages quote it under the same limited-quotation practice as the rest of the site.
• Your source-responsibility statement in the archive is your representation that the material was collected lawfully, in accordance with the publishers' published access rules, and for this purpose only. You are responsible for that collection.

4.6 Acceptance is not publication

An upload does not guarantee review, acceptance, or publication. Every submission must pass deterministic checks, an independent audit, and a final human review before it is published, and we may decline it at any stage without giving reasons.

4.7 Withdrawal

The holder of the withdrawal key may withdraw a published report at any time by uploading a withdrawal archive produced by the toolkit together with the key. Withdrawal removes the report from the public site and records a withdrawal event; the private audit records are kept under section 4.4.

There is no revision procedure. To publish a corrected report, withdraw the existing one and submit a new archive under a new invitation.

An anonymous sponsor holds the same withdrawal right as a named one, but only through the key. If you lose the key and gave no email address, no one can prove the report is yours, and we cannot act on a withdrawal request.

4.8 Our right to remove

We may remove or temporarily withdraw a published report at any time, without waiting for the contributor, for safety, suspected infringement, misleading content, legal requests, or any other reason we consider sufficient. We will tell you if you gave an email address.

5. Human verification by Cloudflare Turnstile

Both forms use Cloudflare Turnstile to tell people from bots. Turnstile processes the technical data it needs for that check, such as your IP address and browser characteristics, under Cloudflare's privacy policy. The site keeps only the result of the check, never the token.

6. Your rights over personal data

What counts as personal data here is small: an email address, a name you entered for credit, and anything personal you chose to type into free-text fields. Reports, suggestions, audit records, third-party text, and status events are not your personal data and are not covered by this section.

You may ask us to:

• tell you what personal data we hold about a suggestion or submission you made;
• correct a public credit line or change it to anonymous;
• delete your email address and name from our records, and remove your credit line from a public page.

How we verify it is you. Write to the contact address from the email address you gave on the form, or, for a submission, include the withdrawal key. If you gave neither, we have no way to confirm that a request comes from the person who submitted, and we will not act on it. This is a verification requirement, not a denial of the right.

What deletion does. We delete the personal fields from the cloud record and from our private working records, and re-publish any affected page without your credit line. We keep an event that records the deletion without your text. We do not delete the suggestion or report itself, published pages, or audit records, because those are not your personal data and are kept as this site's publishing record.

We act on a verified request within 30 days. We do not sell personal data, and we do not share it with anyone except the service providers named in this notice.

Do not put other people's personal data in the forms. Suggestion fields and archives should describe news coverage, not private individuals.

7. Service providers

All four are services of Cloudflare, Inc. We use no other third-party processor for form data.

8. Changes to this notice

The version at the top of this notice is recorded with every form you send, so we always know which text you agreed to. If we change what we collect or how we use it, the version changes and the forms ask for agreement to the new text.