VerifiMind-PEAS — Privacy Policy and Personal Data Protection Notice v2.5
Published and effective: August 6, 2026 (previous: v2.4, August 4, 2026)

1. WHO WE ARE
The hosted service at verifimind.ysenseai.org is operated by Alton Lee Wei Bin
(YSenseAI, an independent project in Malaysia), who is the data controller for
that service. VerifiMind-PEAS is an open-source multi-model AI validation
framework. This notice also covers its Early Adopter (EA) and PILOT programs.

2. CURRENT SERVICE AVAILABILITY
Eight validation and built-in-template tools are currently available. Three
coordination and two custom-template mutation tools are temporarily unavailable
while owner-scoped controls and URL-fetch protections are rebuilt. No paid
services are active. Registration does not create a time-limited entitlement.

3. DATA, SOURCES, AND YOUR CHOICES
We receive data directly from registration, feedback, and validation-tool calls,
and generate limited timestamps, identifiers, and operational records.

Registration requires an email address and acceptance/acknowledgement of the
current Terms and Privacy Policy. A name, feedback, feedback category, and
email-update consent are optional. We generate a pseudonymous UUID, registration
timestamp, tier/cohort label, and consent record. Without the required fields you
cannot register, but the eight active tools remain available anonymously.

Using a validation tool requires concept or prompt content. Additional context,
a UUID, BYOK credentials, and save_to_history are optional. Without the required
prompt the requested validation cannot run. IP addresses and request metadata are
automatically processed for rate limiting, service operation, and abuse/security
investigation. Security logs may be correlated with account or UUID activity only
when reasonably needed for those purposes or legal compliance.

We do not intentionally request or store account passwords, API keys, payment
details, precise-location profiles, or browsing profiles as registration fields.
No paid checkout is active and we never sell personal data. Do not submit
passwords, API keys, credentials, or other secrets in registration, feedback, or
validation prompts. Prompt content you submit is processed as described below.

4. PURPOSES
We process data to provide AI validation, identify registered cohorts, allocate
rate limits, operate dashboards, respond to private requests, process feedback,
send updates when separately consented to, prevent abuse, secure and maintain the
service, and retain evidence of consent and release/security decisions.

5. UUID METADATA AND VALIDATION HISTORY
If you voluntarily supply user_uuid, Cloud Logging records the UUID, tool, tier,
and UTC timestamp for 30 days. A pseudonymous Firestore validation record may
also retain the UUID, validation/tool identifiers, scores, recommendations,
quality indicators, and timestamps. It does not include the submitted concept
name or description. Omitting user_uuid prevents that UUID-linked metadata, but
ordinary IP/request security logs may still be retained.

Separately, save_to_history=true on run_full_trinity (off by default) stores the
full result, including submitted concept text, in a shared, instance-local JSON
history. Raw records are not exposed through the public history resources, but
this store is not owner-scoped. It retains at most the 20 newest opt-in results;
the oldest entries are evicted whenever the store is read or written, and an
instance replacement clears the store. Because eviction also depends on instance
activity and lifetime, no fixed time-based retention period is guaranteed. Leave
save_to_history false for private or sensitive concepts.

Runtime custom-template registration and URL import are disabled. The hosted
service exposes built-in templates only and does not currently accept or retain
custom prompt templates.

6. STORAGE, RETENTION, AND CROSS-BORDER PROCESSING
Account, feedback, and UUID-validation records are stored in Google Cloud
Firestore in the United States (us-central1). Operational logs are held in Google
Cloud Logging for 30 days. EA/PILOT account records are retained for the duration
of membership plus 90 days unless deletion is requested earlier. Identifiable
feedback is reviewed for deletion or anonymisation when no longer needed and no
later than six months after the relevant follow-up closes, subject to a lawful or
documented security/legal hold. Pseudonymous UUID-validation records are included
in the scope of an account deletion request, subject to the same holds.

An accepted deletion request immediately de-identifies the principal account PII.
Remaining personal data is targeted for purge within 7 business days. A legal
obligation or documented security/legal hold may delay or limit deletion; where
permitted, we will explain that limitation to the requester.

Validation prompts and chained agent context are processed outside Malaysia when
hosted inference uses the providers in Section 7. Do not submit personal data or
confidential material unless you have authority to send it to those providers.

7. PROCESSORS AND DISCLOSURE
Google Cloud Platform hosts the service, Firestore, and operational logs and may
process account, feedback, validation-metadata, prompt, and request data needed to
provide that infrastructure.

Google Gemini and Groq are the active hosted AI-inference providers. The selected
provider receives the concept/prompt, optional context, and any prior agent output
needed to generate the requested analysis. Z and CS may use Gemini as a
construction-time fallback when Groq cannot be configured; in-flight provider
failover is currently disabled.

If you choose BYOK, your prompt and ephemeral API key are sent to the provider you
select (Google Gemini, Groq, OpenAI, Anthropic, Cerebras, or Mistral; Ollama may run
locally). Provider processing, location, and retention are also governed by the
selected provider's terms and your provider-account settings. BYOK keys are used
for the call and are not intentionally written to application storage.

We may disclose data when required by law or to investigate and contain abuse or
security incidents. We do not sell personal data and no payment processor receives
registration data while paid checkout is inactive.

8. YOUR RIGHTS AND PRIVATE REQUEST CHANNEL
GET /early-adopters/status/{your-uuid} returns only a limited account-status
summary; it is not a complete personal-data access response. For a private access,
correction, deletion, consent-withdrawal, inquiry, or complaint request, email
alton@ysenseai.org (fallback: creator35lwb@gmail.com) and include your UUID plus
enough information to verify the request. Do not post personal data in a public
GitHub Discussion. General, non-personal questions may use GitHub Discussions.
A postal address for formal private correspondence is available through the same
verified email channel.

You may stop optional UUID analytics by omitting user_uuid, leave
save_to_history=false to avoid the shared full-result history, and withdraw
optional email-update consent. These requests are free of charge.

9. SECURITY AND COOKIES
We use restricted infrastructure access and data minimisation, but no hosted
service can guarantee absolute security. VerifiMind-PEAS does not use tracking
cookies.

10. CONTACT
Private privacy/data requests: alton@ysenseai.org
Fallback email: creator35lwb@gmail.com
General public discussion:
github.com/creator35lwb-web/VerifiMind-PEAS/discussions

11. CHANGES
This v2.5 notice is effective on publication because it adds and enforces a
20-entry cap on opt-in shared history, shortens possible storage, and corrects
service facts; it does not introduce a fee, remove a user right, or shorten an
entitlement. We will notify registered users of
future material adverse changes at least 14 days before they take effect. The
current notice is always available at
verifimind.ysenseai.org/privacy.

12. NOTIS PERLINDUNGAN DATA PERIBADI — BAHASA MALAYSIA
Versi 2.5 — Diterbitkan dan berkuat kuasa pada 6 Ogos 2026

SIAPA KAMI
Perkhidmatan di verifimind.ysenseai.org dikendalikan oleh Alton Lee Wei Bin
(YSenseAI, projek bebas di Malaysia), selaku pengawal data bagi perkhidmatan ini.
Notis ini turut meliputi program Early Adopter (EA) dan PILOT.

DATA, SUMBER DAN PILIHAN ANDA
Kami menerima data terus daripada borang pendaftaran, maklum balas dan panggilan
alat validasi, serta menjana cap masa, UUID samaran dan rekod operasi yang terhad.
Pendaftaran memerlukan alamat e-mel serta penerimaan/pengakuan Terma dan Dasar
Privasi. Nama, maklum balas dan persetujuan menerima e-mel kemas kini adalah
pilihan. Tanpa medan wajib anda tidak boleh mendaftar, tetapi lapan alat aktif
masih boleh digunakan secara tanpa nama.

Penggunaan alat memerlukan kandungan konsep atau arahan. Konteks tambahan, UUID,
kelayakan BYOK dan save_to_history adalah pilihan. Alamat IP dan metadata
permintaan diproses secara automatik untuk had kadar, operasi, keselamatan dan
pencegahan penyalahgunaan. Kami tidak meminta kata laluan akaun, kunci API,
butiran pembayaran atau rahsia sebagai medan pendaftaran. Jangan masukkan rahsia
dalam arahan, pendaftaran atau maklum balas.

TUJUAN PEMPROSESAN
Data digunakan untuk menyediakan validasi AI, mengurus kohort dan had kadar,
mengendalikan papan pemuka, menjawab permintaan peribadi, memproses maklum balas,
menghantar kemas kini dengan persetujuan berasingan, mencegah penyalahgunaan,
menjaga keselamatan perkhidmatan dan menyimpan bukti persetujuan.

UUID DAN SEJARAH VALIDASI
Jika user_uuid diberikan secara sukarela, Cloud Logging menyimpan UUID, alat,
tahap dan cap masa UTC selama 30 hari. Firestore juga boleh menyimpan metadata
validasi samaran seperti skor, cadangan, ID validasi, petunjuk kualiti dan cap
masa, tetapi bukan nama atau penerangan konsep. Tanpa UUID, metadata berkaitan
UUID tidak disimpan; log keselamatan IP/permintaan biasa masih boleh disimpan.

save_to_history=true (lalai: false) menyimpan keputusan penuh termasuk teks
konsep dalam sejarah JSON setempat kepada instans yang dikongsi dan belum
diasingkan mengikut pemilik. Rekod mentah tidak dipaparkan melalui sumber sejarah
awam. Hanya 20 keputusan pilihan masuk yang terbaharu dikekalkan; rekod tertua
dikeluarkan apabila stor dibaca atau ditulis, dan penggantian instans mengosongkan
stor. Tiada tempoh simpanan berasaskan masa yang tetap kerana pengeluaran juga
bergantung pada aktiviti dan jangka hayat instans. Kekalkan false untuk konsep
peribadi atau sensitif.

STORAN, TEMPOH SIMPANAN DAN PEMINDAHAN RENTAS SEMPADAN
Rekod akaun, maklum balas dan metadata validasi disimpan dalam Google Cloud
Firestore di Amerika Syarikat (us-central1). Log operasi disimpan 30 hari. Rekod
EA/PILOT disimpan sepanjang keahlian tambah 90 hari kecuali pemadaman diminta
lebih awal. Maklum balas yang boleh dikenal pasti disemak untuk pemadaman atau
penyahpengenalan selewat-lewatnya enam bulan selepas tindakan susulan tamat,
tertakluk kepada undang-undang atau penahanan keselamatan/undang-undang yang
didokumenkan.

Apabila permintaan pemadaman diterima, maklumat peribadi utama akaun dinyahkenal
pasti dengan segera dan baki data peribadi disasarkan untuk dipadam dalam 7 hari
bekerja. Kewajipan undang-undang atau penahanan yang didokumenkan boleh mengehadkan
atau melambatkan pemadaman.

PEMPROSES DAN PENZAHIRAN
Google Cloud Platform menyediakan hos, Firestore dan log. Google Gemini dan Groq
ialah penyedia inferens AI hos yang aktif; penyedia terpilih menerima arahan,
konsep, konteks pilihan dan output ejen terdahulu yang diperlukan. Z dan CS boleh
menggunakan Gemini sebagai sandaran semasa pembinaan jika Groq tidak dapat
dikonfigurasi; pertukaran penyedia semasa permintaan sedang berjalan tidak aktif.

Jika anda memilih BYOK, arahan dan kunci API sementara dihantar kepada penyedia
pilihan anda (Google Gemini, Groq, OpenAI, Anthropic, Cerebras atau Mistral;
Ollama boleh berjalan secara tempatan). Pemprosesan, lokasi dan tempoh simpanan
penyedia turut tertakluk kepada terma penyedia dan tetapan akaun anda. Data juga
boleh dizahirkan apabila diwajibkan undang-undang atau untuk menyiasat insiden
keselamatan. Data peribadi tidak dijual.

HAK DAN SALURAN PERMINTAAN PERIBADI
GET /early-adopters/status/{uuid} hanya memberikan ringkasan status dan bukan
jawapan akses data yang lengkap. Untuk akses, pembetulan, pemadaman, penarikan
persetujuan, pertanyaan atau aduan secara peribadi, e-mel alton@ysenseai.org
(sandaran: creator35lwb@gmail.com) dengan UUID dan maklumat yang mencukupi untuk
pengesahan. Jangan siarkan data peribadi dalam GitHub Discussion awam. Alamat pos
untuk surat-menyurat rasmi boleh diperoleh melalui saluran e-mel yang disahkan.
Anda boleh menghentikan analitik UUID dengan tidak memberikan user_uuid,
mengekalkan save_to_history=false, dan menarik persetujuan e-mel pilihan.

KESELAMATAN, KUKI DAN PERUBAHAN
Kami menggunakan akses infrastruktur terhad dan peminimuman data, tetapi tiada
perkhidmatan hos dapat menjamin keselamatan mutlak. Kami tidak menggunakan kuki
penjejakan. Versi 2.5 menguatkuasakan had 20 rekod untuk sejarah pilihan masuk,
memendekkan kemungkinan penyimpanan dan membetulkan fakta tanpa mengurangkan hak
atau memperkenalkan bayaran. Pengguna berdaftar akan dimaklumkan sekurang-kurangnya
14 hari sebelum perubahan material yang memudaratkan pada masa hadapan berkuat kuasa.