Privacy Policy
Last updated: October 24, 2026
PixJelly at https://pixjelly.com (the "Service") is built and operated by an independent developer ("we", "us" or "our"). This Privacy Policy explains what personal information we collect when you use the Service, how we use and share it, how long we keep it, and the rights you have.
By using the Service you agree to this Policy. If we make material changes, we will tell you in advance as described in Section 13.
1. Data controller
- Controller: the independent developer who operates PixJelly.
- Privacy contact: hello@pixjelly.com
- Data Protection Officer: not applicable.
2. Personal information we collect
2.1 Information you provide
- Google sign-in: when you sign in with Google we receive your Google account ID, email address, name and profile picture. Our database stores only the Google account ID. Your name, email and picture are kept in an encrypted session cookie on your device so we can show who is signed in; we do not store them in our database.
- Images and instructions: the photos you add, the marks and text you draw on them, and the prompts you write. What happens to them depends on the tool — see Section 3 and Section 5.
- Feedback: the message you send through the in-app Feedback form (up to 500 characters), an optional contact email, and the page you sent it from. Our replies are stored with it.
- Payment information: when payments are enabled, purchases are processed by our merchant of record, Waffo Pancake. We receive the order and transaction details (product, amount, status, order ID) but never your full card number — card data is handled by Waffo Pancake and is not stored on our servers.
2.2 Information we collect automatically
- Credit ledger: your credit balance and a record of each generation request (request ID, model, cost, status and time) so that credits are charged correctly and refunded when a request fails. The images themselves are not part of this record.
- First visit source: the first time you visit, we store in a cookie where you came from — the referring website’s domain (for example chatgpt.com or google.com), any utm_source / utm_medium / utm_campaign tags in the link, the page you landed on and the time. We keep only the domain and the page path, never full URLs, and no other link parameters. If you later sign in, this source is saved with your Google account ID.
- Free AI edits without signing in: if you use AI redraw or the annotate editor without signing in, we count uses per day using a keyed hash (HMAC) of your IP address and of a random browser ID stored in your browser. We never store your raw IP address for this purpose. While you wait in the free queue we keep a record of your place: the two hashed keys, the tool, and when you joined, were last seen and started.
- Bot check: before a free AI edit without signing in, Cloudflare Turnstile runs a check in your browser and we confirm the result with Cloudflare. We keep whether the check passed, any error codes and the hashed IP key; we do not keep the check token or your raw IP address.
- Server and security logs: our hosting provider records technical request data such as IP address, user agent, time and the page or API requested. We also limit how many generation requests one IP address can make per hour, using a short-lived in-memory counter.
- Analytics: we use Vercel Web Analytics to count page views, referrers and campaign tags. It does not use cookies.
- Safety check records: every request that runs on our servers is checked by an automated content safety system (Section 3). For each check we keep a record with a one-way hash of the prompt or image, the safety scores, the decision, the time and, if you are signed in, your Google account ID. We do not keep the prompt text or the image in these records.
3. How we use your information
| Purpose | Legal basis |
|---|---|
| Providing the Service: running edits, generations and background removal you request; keeping you signed in | Performance of a contract |
| Credits and billing: charging credits, refunding failed requests, processing purchases | Performance of a contract |
| Answering feedback and support requests | Performance of a contract / legitimate interests |
| Preventing abuse: rate limits, the daily free allowance, the free queue and the bot check (Cloudflare Turnstile) | Legitimate interests |
| Content safety: checking prompts, uploaded photos and results against our content rules before a request starts and before a result is delivered | Legitimate interests / legal obligation |
| Understanding which channels bring visitors and customers (first visit source, analytics) | Legitimate interests |
| Complying with law and responding to lawful requests | Legal obligation |
Some tools never send your photo to us. Free background removal, White Background and Quick erase run entirely in your browser: the model files are downloaded to your device and your photo stays there.
Other tools need a server. Image editing, Text to Image, Image to Image, AI redraw and the Standard and HD background removal tiers send your image and instructions to the providers listed in Section 5 to produce the result.
Before a server-side request starts, we send your prompt (including any text you add to the image) and each photo you upload to OpenAI’s moderation model; before a result is delivered, we send the result image too. If something does not pass, the request is not started, or the result is not delivered and its credits are refunded.
We do not use your images, prompts or results to train AI models. What our providers say about training (their published terms, checked October 24, 2026):
- KIE, which runs the image models on our behalf, does not publish a training policy. Its terms give it a licence to use submitted content to operate and improve its service, and it does not say which model provider terms apply to the requests it sends.
- OpenAI does not train on data sent to its moderation endpoint, which we call directly.
- For API customers, OpenAI does not train on API data by default; Recraft does not use API content for training; ByteDance’s BytePlus ModelArk does not use customer content to train its models without the customer’s authorisation; and Google does not use content sent to its paid Gemini API to improve its products. For models run through KIE, KIE is the customer of these providers, not us.
- Photoroom’s privacy policy and API documentation say images sent through its API are not used for training. Its security page says images from self-serve API accounts may be used for training unless the account opts out in its settings.
- Vercel, our hosting provider, does not use our data for AI training on our plan.
- Cloudflare uses the signals collected by Turnstile (such as IP address, TLS fingerprint and user agent, not your images) to improve its bot detection, acting as an independent controller for that purpose.
We may produce aggregated or anonymized statistics that cannot be linked to you.
6. Data security
- All traffic to the Service is encrypted with HTTPS / TLS.
- Sign-in sessions are stored in encrypted, HTTP-only cookies. We never see or store your Google password.
- Provider credentials stay on our servers and are never exposed to the browser. Database access is limited to our server, with row-level security enabled and public access revoked.
- Free-use counters store keyed hashes, not raw IP addresses.
If a security incident affects your rights, we will notify you and the relevant authorities within 72 hours of discovery where required by law.
7. Data retention
| Data | How long we keep it | Then |
|---|---|---|
| Images you upload and results generated on our servers | We do not store them in our database or file storage. Results are available to you through a link that expires after 24 hours. Copies held by providers follow Section 5. | — |
| Images processed on your device | Never leave your device | — |
| Account record (Google account ID, credit balance) | While your account is active; deleted within 30 days of a deletion request | Deleted |
| Credit ledger and transaction records | As long as tax and accounting rules require, up to 7 years | Deleted |
| Feedback and our replies | 2 years | Deleted |
| First visit source | Cookie: 180 days. Saved with your account while it is active | Deleted with your account |
| Free AI edit counters (hashed) | 30 days | Deleted |
| Free queue places and bot check results (hashed keys, no images) | 30 days | Deleted |
| Safety check records (hashes and scores, no prompts or images) | 180 days | Deleted |
| Server logs held by our hosting provider | 1 day | Deleted by the provider |
8. Your data rights
To exercise any of the rights below, email hello@pixjelly.com from the address linked to your Google account. We respond within 30 days.
| Right | What it means |
|---|---|
| Right to be informed | Know what we collect and how we use it — this Policy |
| Right of access | Get a copy of the personal information we hold about you |
| Right to rectification | Correct inaccurate information |
| Right to erasure | Ask us to delete your account and data, subject to legal retention duties |
| Right to restrict processing | Ask us to pause certain processing |
| Right to data portability | Receive your data in a machine-readable format |
| Right to object | Object to processing based on legitimate interests |
| Right to withdraw consent | Withdraw consent where we rely on it |
You can also lodge a complaint with your local data protection authority.
9. Marketing and opt-out
We do not send marketing emails or messages. If we ever start, we will ask for your consent first, include an unsubscribe link in every message, and keep essential service notices (billing, security, policy changes) separate.
10. International data transfers
Our database is hosted by Supabase in India (AWS region ap-south-1). Our website and server functions are hosted by Vercel, with server functions in Washington, D.C., United States (Vercel region iad1). The providers in Section 5 process data in these locations, as they publish them: KIE may transfer data to the United States and other countries; Google may process data in any country where it has facilities; OpenAI does not specify a location by default; BytePlus (Seedream) processes data in Malaysia, Indonesia and/or the EU/EEA; Recraft processes data in the United States and the countries of its vendors; Photoroom processes data in the United States; Cloudflare runs Turnstile on its global network.
Where personal information is transferred across borders, we rely on the data processing terms of our providers. Vercel’s Data Processing Addendum, which includes the EU Standard Contractual Clauses and the UK Addendum, applies to our plan, and OpenAI’s Data Processing Addendum, with the same clauses, covers the moderation checks we send to OpenAI directly. Google, OpenAI, BytePlus and Recraft publish data processing agreements that include the Standard Contractual Clauses; for models run through KIE they apply between those providers and KIE. Cloudflare publishes a customer data processing agreement that includes the Standard Contractual Clauses and the EU–US Data Privacy Framework. KIE does not publish a data processing agreement, and Photoroom offers one only to enterprise customers. If you have questions about a specific transfer, contact us.
11. Children’s privacy
The Service is intended for users aged 18 and above. We do not knowingly collect personal information from children below that age. If you believe a child has given us information, contact us and we will delete it promptly.
12. Third-party links and services
The Service links to, and integrates with, third-party services such as Google sign-in and the Waffo Pancake checkout. This Policy covers only information we collect. Please review those services’ own privacy policies.
13. Changes to this Policy
For material changes we will give at least 30 days’ notice by a notice on the site or by email, and update the date at the top of this page. Continued use after the effective date means you accept the updated Policy.
14. Contact us
- Privacy and support: hello@pixjelly.com
- We reply to emails within 2 business days.
This Privacy Policy is provided for transparency and does not constitute legal advice.