Frequently Asked Questions
General
What is Sourcemark?
Sourcemark is a consent registry for AI training data. It lets creators declare whether their work is available for AI training, and gives AI companies a registry to check before they train. Think of it as a clear, machine-readable layer of consent between the people who make creative work and the companies building AI models.
Who is Sourcemark for?
Three audiences. Individual creators — photographers, illustrators, designers, filmmakers, writers — who want to make their AI training preferences known. Rights-holding organisations like studios and agencies managing large libraries of work. And AI companies that want a structured, auditable way to check consent before including files in training datasets.
Is Sourcemark anti-AI?
No. Sourcemark is not about blocking AI training. It is about making creator choice visible. Some creators will say their work is not available for AI training. Others will make it available under a licence. Both positions are equally valid, and Sourcemark records them the same way.
How is Sourcemark different from robots.txt?
Robots.txt is a loose signal that applies to an entire website, has no timestamp, and is routinely ignored. A Sourcemark record is tied to a specific file, timestamped at the moment of creation, machine-readable, and backed by an audit trail of any changes. It works at the file level, not the domain level.
Is Sourcemark a licensing marketplace?
No. Sourcemark records your consent declaration and provides a contact pathway so interested parties can reach you. It does not handle licence negotiations, payments, or rights management.
Is Sourcemark a copyright registry?
No. Sourcemark records that you made a declaration about a file. It does not certify or prove ownership. It is a consent registry, not a copyright registry.
Does the absence of a Sourcemark record mean consent?
No. If a file has no record in the registry, that means no one has made a declaration about it. Absence is not consent. AI companies should not treat an unregistered file as permission to train.
What content types does Sourcemark support?
Images (JPEG, PNG, WebP, TIFF), PDFs, and video files. If your creative work lives in these formats, it can be sourcemarked.
How long does it take to create a Sourcemark?
A few minutes for a single file. You select your file, set your preferences, and submit. The fingerprinting happens locally in your browser and takes seconds even for large files.
What browsers are supported?
Sourcemark works in all modern browsers — Chrome, Firefox, Safari, and Edge. No plugins or extensions are required.
For Creators
What choices can I make about my work?
Two options. You can declare your work not available for AI training, or you can make it available by licence. If you choose available by licence, you can include contact details so interested parties know how to reach you.
Does sourcemarking my work stop it from being used without permission?
No. No tool can prevent a determined actor from downloading publicly available files. Sourcemark works at the consent layer, not the access layer. What it does is create a checkable record of your declared preference — so if your work is used, there is a clear, timestamped record of what you said.
Does Sourcemark prove I own my work?
No. A Sourcemark record states that you made a declaration about a file at a specific point in time. It does not prove creation or ownership. It is a consent declaration, not a certificate of authorship.
Do you store my files?
No. When you sourcemark a file, the fingerprinting happens entirely in your browser. Only the fingerprints are sent to the registry. Your original file never leaves your device.
What is a fingerprint?
Sourcemark generates two types. A SHA-256 hash is an exact fingerprint — change a single byte and it changes. A perceptual fingerprint captures the visual essence of an image, so it can match files even after resizing, recompression, or minor edits. Together, they cover both exact and near-duplicate matching.
What is the verification page?
Every Sourcemark record has a unique public page showing the file's consent status, when the record was created, and any changes over time. Anyone can view it without an account. You can share it with clients, include it in your portfolio, or link to it from your website.
Can I change my mind after creating a record?
Yes. You can update your AI training preference at any time. Sourcemark preserves the full history of changes, so the timeline of your decisions is always intact. The current status is what AI companies see when they query the registry.
Can I add a licensing representative?
Yes. If you want enquiries about your work to go through an agent, studio, or other representative, you can include their contact details as the licensing route on your record.
What about work I've already published online?
You can still sourcemark it. Creating a record is not limited to new work. If your files are already circulating on the internet, sourcemarking them creates a declaration from this point forward.
Does my Sourcemark record apply to past use of my work?
A record is timestamped from the moment you create it. It does not retroactively cover prior use. But it establishes a clear, dated declaration that applies from that point onward.
Do I need C2PA content credentials?
No. C2PA is entirely optional. If your camera or software supports it, Sourcemark detects and preserves those credentials automatically. If not, standard fingerprinting works for any supported file.
Is Sourcemark free?
Yes for individuals. You can create records, manage your library, and share verification links at no cost. Paid plans are available for studios and organisations that need batch operations, team management, and advanced features.
Is my data secure?
Yes. Your files never leave your device — only fingerprints are stored. The registry holds the minimum information needed to answer consent queries: who declared, what file, when, and what they said.
Can I export my records?
This is a planned feature. We're working on export options so you can download your full record history in standard formats.
For AI Companies
How do I access the Sourcemark registry?
Register for an account and apply for API access. API access is available on paid enterprise plans and gives you programmatic access to query the registry at scale.
What does a query return?
A structured response containing the consent status (not available or available by licence), licensing contact information if provided, the timestamp of the original declaration, and the full history of any changes.
Can I run bulk queries?
Yes. The API is designed for batch lookups. You can query multiple fingerprints in a single request, which makes it practical to integrate into training data pipelines.
What fingerprint formats are supported?
SHA-256 for exact matching and perceptual hashes for visual similarity matching. You can query using either or both.
Does a Sourcemark record guarantee compliance?
No. Sourcemark provides structured, timestamped consent records. It does not offer legal advice or guarantee regulatory compliance in any jurisdiction. It gives you a clear, auditable record of what a creator declared and when.
What if there is no match for a file?
No match does not mean consent. It means no one has made a declaration about that file in the registry. The absence of a record should not be treated as permission to include the file in training data.
Can I integrate Sourcemark into my training pipeline?
Yes. The API is designed for exactly this. You can add a consent check step to your data ingestion pipeline that queries the registry before any file enters your training set.
Want to go deeper? Read our guide for AI companies or learn what Sourcemark is.