Healthyasat

Privacy Notice

Last updated: 16 October 2026

Healthyasat is a health-misinformation detector for Taglish (Tagalog-English) content. This notice describes what the Healthyasat browser extension and the fact-checking service at api.healthyasat.fit handle, where that data is stored, and how to reach us. It applies to the extension and to this website.

Data kept on your device

Your analysis history, settings (language, theme, detection toggles), the local rate-limit state, and a random device client ID are stored only on your device, in Chrome's local extension storage. We keep no copy on our servers, no account is required, and you can clear this data at any time from the extension's settings.

Data sent to the fact-checking API

When you ask the extension to check a claim (a text selection or a YouTube transcript), it sends only what is needed to fact-check that request:

  • the selected text, or the YouTube transcript excerpts being checked;
  • the URL, platform, and, where available, author or channel metadata of the page or video;
  • the content location (which part of the page or video the claim came from);
  • a random, device-generated client ID used to group rate limits — it is not tied to your identity;
  • your language setting.

The server also sees the source IP address of the request as part of normal internet operation. It is used only to enforce rate limits and prevent abuse: the stored rate-limit state is a one-way hash of the IP and client ID, never the raw address, and the IP is never stored together with the analysis content.

How the data is used

Requests are used only to produce a fact-check verdict (accurate, inaccurate, or unverified) and to return it to you. We do not use them for advertising, profiling, or analytics, and we do not sell or share them.

Server-side storage

To make fact-checking faster and consistent, and to support source-reputation scoring, the service stores the following for each analysed claim:

  • the source URL and the claim text;
  • the returned verdict and its supporting detail;
  • an anonymous timestamp.

These records contain no account, name, email, or other personally identifiable information. They are stored in our Supabase Postgres database (region: Seoul, ap-northeast-2).

Third parties and sub-processors

  • DeepSeek — the only third-party model vendor. The gatekeeper (is this a health claim?), the Taglish-to-English translation, and the verdict generation all run on the DeepSeek API. Only the content described above is sent.
  • Supabase — managed Postgres for the server-side records.
  • Qdrant Cloud — hosts the health-myth fact dataset used for retrieval (region: Sydney, australia-southeast1). This host stores the fact dataset only, not user requests.
  • Cloudflare — DNS, web application firewall, and TLS in front of our servers.

No other third-party AI or model vendor processes extension requests. The admin portal's dataset tooling writes to the same fact dataset; no hosted embedding vendor is involved.

Your rights under the Data Privacy Act of 2012 (RA 10173)

We collect no personally identifiable information, so there is no personal profile to correct or export. You still have the right to be informed, to object, and to request deletion of any server-stored analysis records we hold (URLs, claim texts, verdicts, and anonymous timestamps).

Changes

We may update this notice as the system changes. The date at the top always reflects the current version.

Contact

For privacy questions or to request deletion of server-stored data, email us at [email protected].