Who is responsible
Material Evidence is operated by Fareed Sayegh in Las Vegas, Nevada. Privacy contact: [[NEEDS INPUT: privacy contact email]]. Mailing address: [[NEEDS INPUT: complete business mailing address, if one should be published or is legally required]].
Information you choose to submit
- Finder entries. Clothing descriptions and product links are used to classify an item and produce alternatives.
- Garment photos. A photo is used to identify the garment category and visible style attributes. Do not upload a photo of a person. Ordinary classification does not retain the upload. If you separately opt in on the results page, a small reference copy is saved privately as described below.
- Manual fiber composition. A composition you type is used in the browser to compare synthetic content and rank alternatives.
- Catalog requests. If you ask us to source an item, the request text and selected category are stored in the private curation queue.
- Internal discovery criteria. Authorized administrators may submit a category, audience, and criteria to find products for review.
Information created when you use the site
- Finder results. The service creates category, confidence, and style-attribute results from a description, link, or photo.
- Product-page evidence. For a pasted link, the service may store the normalized URL, parsed product name, composition, finish claims, price, image references, and analysis result temporarily in a server cache.
- Anonymous outbound-click events. When you open a retailer link, the site records the item slug or Tier 2 URL, brand, category, result tier, arrival source, and timestamp. It does not add a name, account identifier, or IP address to the click event.
- Rate-limit records. The service derives a truncated cryptographic hash from the request IP address and an endpoint label. It stores the hash, a count, and an expiry. The raw IP address is not stored in that table.
- Live discovery records. Tier 2 products shown repeatedly are logged with product-page facts and a surface count so candidates can be reviewed later.
Optional clothing research
The separate, unticked research checkbox on the browse page allows Material Evidence to save structured garment features, search outcomes, and any product link or garment photo attached for private catalog and future clothing-design research. Searching works without this choice. Consent applies only to new live searches started while checked, not a search already running or finished. Earlier consent to counts-only research does not authorize reference storage.
The aggregate table stores a fixed category, clothing audience, recognized design-feature labels, input method, UTC calendar day, result and checking-failure counts, treatment-claim counts, a random record ID, and consent version. A separate reference record may contain a normalized HTTPS product link, a private image identifier, a consent timestamp, and a hashed removal code. Tracking parameters and fragments are removed from links while selected product identifiers are retained. We do not save raw search descriptions, original filenames, account identifiers, or IP addresses in these research records.
For an attached photo, the browser creates a small JPEG copy; the server validates its format and dimensions and removes embedded metadata and comments before private object storage. The original uploaded file is not retained for research. Only submit garment-only images you have permission to share, without people, children, faces, or personal details. Photo pixels and product links can still contain identifying information, so we do not describe these references as anonymous.
A photo may be held briefly in browser memory to carry it from the finder to the results page. No photo is written to cookies, localStorage, or sessionStorage. Reloading loses that draft. No research copy is sent for storage unless you opt in. No additional model call is made for research.
Only the authorized administrator can view saved references. Photos are served through an authenticated, non-cached route, never a public image URL. Research remains separate from the product-analysis cache, retailer-result queue, and security records described elsewhere in this notice.
References and aggregate records are excluded from reports after 180 calendar days, counting the submission day. Expired records and image objects are cleaned up in batches on subsequent research writes or administrator visits; physical deletion can occur later if there is no activity or storage is unavailable. Provider backup and log retention remain subject to the provider details below.
Unchecking the option stops participation in new searches. For each attempted reference submission, you receive a private reference-removal link. Keep it: anyone holding that link can remove that reference without an account. Removal hides the reference immediately and deletes the stored image; failed storage deletions remain hidden and can be retried. Only structured garment counts remain. Deletion receipts containing only a hashed removal code and timestamp are retained for up to 180 days before cleanup, preventing delayed submissions from restoring a removed reference. The administrator can also remove a reference. Existing catalog requests are not retroactively repurposed for clothing-line research.
Purposes
We use the information above to return clothing alternatives, analyze a product page you provide, prevent abusive or excessive requests, measure anonymous outbound interest, improve the vetted catalog, and protect the service. Separately opted-in research helps us identify garments to investigate for a possible future clothing line. Search gaps are not treated as proof that no alternative exists.
Legal bases for EU and UK visitors
Whether EU or UK data-protection law applies depends in part on whether Material Evidence intentionally offers services to or monitors people there, not merely whether the site can be opened there. If those laws apply, the likely bases are:
- Consent for descriptions, product links, garment photos, manual composition, catalog requests, and internal discovery criteria submitted through a form.
- Legitimate interests for proportionate service security, short-lived rate limiting, and first-party aggregate measurement of retailer-link interest. The tradeoff is that this basis requires a documented balancing assessment and a way to object where the law provides one.
- Providing the service you requested may also support processing needed to return the finder result. Whether this is treated as performance of a contract is a legal judgment that depends on how the free service is offered.
Recipients
- OpenAI. Text descriptions, garment photos, product-page images, live-search criteria, and internal discovery criteria may be sent to the OpenAI API to classify or search. API requests set
store: false. OpenAI states that API data is not used to train its models unless the account opts in, and that default abuse-monitoring logs may retain request and response content for up to 30 days. See OpenAI API data controls. - OpenAI Sites and Cloudflare infrastructure. The application is configured for OpenAI Sites hosting and Cloudflare Workers and D1. They process requests and store the application records described above. [[NEEDS INPUT: confirm the contracted hosting entities, data locations, host-log contents, and host-log retention]].
- Cloudflare DNS resolver. When the server checks a product hostname before fetching it, it may send the hostname to Cloudflare's DNS-over-HTTPS resolver.
- Retailers. When you choose an outbound product link, your browser connects to that retailer. The retailer receives ordinary connection information under its own privacy notice. Product links are marked
noreferrer, so the application asks the browser not to send the Material Evidence page as the referrer.
Retention
- Verified pasted-link analysis is cached for up to 24 hours. Category-only results are cached for up to 30 minutes. Results with a style-analysis failure are cached for up to 5 minutes.
- Rate-limit hashes expire after the applicable one-minute, five-minute, or one-hour window and expired rows are periodically deleted.
- OpenAI's default abuse-monitoring retention is described above.
- Catalog requests: [[NEEDS INPUT: retention period for gap requests]].
- Anonymous click events: [[NEEDS INPUT: retention period for outbound-click events]].
- Tier 2 discovery records: [[NEEDS INPUT: retention period for live discovery records]].
- Approved catalog and internal candidate records: [[NEEDS INPUT: retention period or deletion rule for curation records]].
- Hosting and security logs: [[NEEDS INPUT: retention period and fields recorded by the hosting provider]].
Private curator accounts
The invite-only curation dashboard uses the existing Material Evidence account system, Supabase Auth, to send and verify email sign-in codes. Resend delivers these authentication emails through the configured email service. Only explicitly authorized, verified accounts can access drafts or publish products; an ordinary app account is not curation permission.
Sign-in sends the email address and the entered one-time code to Supabase for authentication. The website stores a server-side hash of the short-lived session token, the curator account ID and an expiry. A first-party HttpOnly cookie holds the session token for at most one hour. Refresh tokens are not retained. Signing out revokes that dashboard session; expired session records are excluded immediately and deleted on a subsequent successful sign-in. Codes and raw tokens are not written to application logs or the session database.
Catalog edits record the acting curator account ID, item ID, action and timestamp for accountability. Private brand notes and unpublished drafts remain separate from the public catalog. Review evidence attached to a published item is public. Curation and audit-record retention: [[NEEDS INPUT: retention period or deletion rule for curator drafts, notes and audit events]]. Supabase and Resend provider log retention and processing locations: [[NEEDS INPUT: confirm authentication and email-service retention and processing locations]]. These records support requested dashboard access and service security; they are not marketing subscriptions.
Cookies and browser storage
The public shopping experience does not require a cookie. The private curation dashboard sets an essential sign-in cookie for authorized curators. It does not put codes or session tokens in localStorage or sessionStorage, or load analytics, advertising pixels, or marketing tags. See the cookie notice.
Your choices and rights
You can choose not to submit a description, link, photo, composition, or catalog request and can browse the vetted catalog instead. Depending on where you live, you may have rights to access, correct, delete, restrict, object to, or receive a copy of personal data, withdraw consent for future processing, and complain to a regulator. To exercise a right, contact [[NEEDS INPUT: privacy contact email]]. We may need enough information to locate the record and verify the request, but we will not ask for more than is reasonably needed.
International processing
The service is operated from the United States and its providers may process information in the United States or other countries. If EU or UK transfer rules apply, [[NEEDS INPUT: confirm the provider contract, transfer mechanism, and whether an EU or UK representative is required]].
Children
The kids catalog is for adults looking for children's clothing. The service is not designed for children to submit information themselves. [[NEEDS INPUT: minimum permitted user age and the process for deleting information submitted by a child]].
Changes
We may update this notice when the service or its providers change. The date at the top will show the latest revision. Material changes should receive a more prominent notice before they take effect.