Privacidad
Esta política está disponible en inglés.
We keep visit data, public account details, saved models and email addresses you provide.
This policy covers our website, apps and ChatGPT connector. Hugging Bay is operated by Barney Wohl. Contact barney@huggingbay.xyz about access, correction or deletion.
Data, sharing and retention
Cloudflare carries and protects traffic. Render runs our website and connector. Google Cloud stores service records and backups and delivers model files. They process the data needed to provide those services. The table identifies other recipients and public disclosures. Retention below concerns our service records; logs and backups are listed separately.
| Data | Purpose | Shared with whom and why | Retention |
|---|---|---|---|
| Visits and usage | Pages, browser details, referrer, approximate country and visitor buckets help us count use and diagnose problems. | Cloudflare, Render and Google Cloud for delivery, security and storage. Public traffic reports show aggregates. | Raw visit records are kept for up to 90 days; access records for 30 days. Anonymous aggregates may remain indefinitely. The optional hb_visitor cookie lasts up to 13 months (390 days). |
| Public account details and saved models | Your public username, display name and profile picture identify your account. Saved models, searches and watchlists follow your account. | Your sign-in provider authenticates you; Render and Google Cloud process the account and saved lists. Public publisher profiles and material you publish are visible to others. | Accounts are kept while active and deleted within 30 days of closure. Saved lists remain until you remove them or close your account. |
| Email addresses and messages | We use addresses you provide for requested notifications, account access or support. | Resend processes addresses and messages to send email. Render and Google Cloud store account and notification records. Support messages are available to the operator to answer you. | Notification addresses remain until unsubscribe, account deletion or earlier notification settlement. Unconfirmed subscriptions expire after 7 days. Support emails are kept for up to 2 years. |
| Downloads and app diagnostics | Selected file, outcome, country, IP-derived bucket and app details help deliver downloads and fix failures. Desktop events may include a resettable install ID and hardware class. | Cloudflare, Render and Google Cloud for delivery and diagnostics. App events can be written to service logs. | Download records and app events are deleted after 12 months. Private model requests from the iPhone app are also deleted after 12 months. |
| ChatGPT tool inputs and results | Search words, model IDs, tasks, filters and hardware details produce model recommendations, file links and hashes. | Cloudflare and Render process requests; Google Cloud supplies catalog records. Results are returned to OpenAI for your ChatGPT conversation. | Search and query logs are deleted after 90 days. Cached answers may be reused for up to 30 minutes and are removed on eviction or process restart. Technical metrics are described below. |
| Connector technical metrics | Keyed identifiers, session counts and tool outcomes help us measure reliability and use. | Render and Google Cloud process metrics for our operators. Hashing does not make identifiers anonymous. | Cohort identifiers are deleted after 30 UTC days and daily cohort metrics after 90 UTC days. Legacy search and query records are deleted after 90 days. Anonymous usage totals may remain indefinitely. |
| Confirmed model requests | The public model repo you confirm records demand and lets you check request status. Internal keyed references prevent duplicate votes. | Render and Google Cloud process public requests and duplicate checks. The ChatGPT connector uses public-only V2 requests: the repo and anonymous demand counts are public; private task text is not retained as a request. Internal keyed references are not public or anonymous. | Public repo and anonymous demand totals may remain indefinitely. Keyed duplicate-check references remain with the request record. Private iPhone model requests follow the 12-month limit above. |
| Payments | Billing account details and transaction references support paid services, where used. | Stripe processes payments. We do not receive your full card number. | We keep billing records for 7 years for tax purposes. Stripe retains its payment records under its own legal and regulatory requirements. We do not store card numbers. |
| Service logs and backups | Network and diagnostic records support security and recovery. Backups may contain the service records above. | Cloudflare, Render and Google Cloud provide traffic protection, logs and recovery storage. | Provider logs follow their default schedules, described below. Database recovery backups are kept for 14 days; manual backups are deleted within 90 days of creation. Deleted records can remain in those backups until expiry. |
Render dashboard logs last 7 days on Hobby, 14 on Pro and 30 on Scale or Enterprise. Google Cloud default logs last 30 days and required administrative audit logs 400 days. Cloudflare does not retain Logpull HTTP logs by default; if enabled, they are available for 3 to 7 days. Other Cloudflare security and analytics records follow the product and plan schedule. Resend keeps email and log data for 30 days on Free, Pro and Scale; enterprise retention follows its contract. Resend backups last 7 days. These are provider schedules, separate from our service-record limits.
ChatGPT connector
The connector uses public model tools without a Hugging Bay sign-in. We receive the arguments ChatGPT sends to a tool, not automatic access to your entire chat history. Do not include secrets or unrelated personal information in searches or tasks. Results contain public model and publisher details, download evidence and, when requested, confirmation or status information.
A public missing-model request requires a separate confirmation before it is recorded. Its preview states that the repo and demand become public, and its confirmation and status are returned to ChatGPT. We do not keep a private task journal for new ChatGPT V2 requests. OpenAI handles your conversation and tool results under its own terms and privacy policy. Following a publisher or download link sends a new request to that destination.
Your choices
You can remove saved entries, unsubscribe from notification emails and ask us about access or deletion at the contact above. Declining visit counting, Do Not Track or Global Privacy Control suppresses our supported visit and connector collectors. These choices do not turn off providers' security logs or erase existing records. We do not sell personal data or share it for advertising. Our service providers are Render, Google Cloud, Cloudflare, Resend, Stripe and Hugging Face for sign-in. You can ask the operator at barney@huggingbay.xyz to handle access or deletion; billing records retained for tax follow the period above.
Hugging Bay for iPhone
Your chats, drafts, answers and downloaded models stay on your phone. We never receive them.
To find and download models, the app sends us your search words, the model you pick, and any model name you ask us to add. Search words are used to answer your search and are not linked to you.
For each download we record the file, whether it finished, the app type, an approximate country, and a keyed hash of your IP address that we cannot turn back into an address. We use this to deliver files, fix problems and count downloads.
No tracking, no ads, and no third-party SDKs in the app. We do not sell your data.
The iPhone app is for adults (rated 18+).
Updated 9 November 2026. © 2026 Hugging Bay.