Keep strong records of your original work, make your terms visible, use the platform controls available to you, consider registration where it fits, and evaluate technical tools by the specific threat they address. None of these steps is a complete shield. Together, they create better evidence, clearer boundaries, and more options when something goes wrong.
1. Keep the originals and the trail.
Save source files, drafts, project files, export history, dated notes, contracts, and publication records. Keep a copy outside the platform where the work appears. A clear creation trail will not prevent copying, but it can help establish what you made and when.
For commissioned work, put ownership and permitted uses in writing. If AI training, style imitation, voice cloning, or synthetic derivatives matter to you, name them instead of relying on a general understanding.
2. Make your terms visible.
Add a plain-language rights or licensing page to your site. State whether the work may be reproduced, licensed, used in datasets, or used to train generative systems. Put a short notice near the work when practical and include a way to ask for permission.
A notice does not physically stop a scraper, and it does not decide every legal question. It removes the excuse that your position was impossible to find.
3. Review platform terms before posting.
Social and portfolio platforms can change how uploaded work may be processed, licensed, or used. Check the current terms and privacy controls instead of relying on an old screenshot or a secondhand summary. Look for settings related to model training, data sharing, public visibility, downloads, and third-party access.
If a platform offers an opt-out, use it and keep a record. An opt-out is only as strong as the system enforcing it, but it is still a clear expression of your choice.
4. Consider copyright registration where it fits.
Copyright rules vary by country. In the United States, the Copyright Office registration portal explains the registration process and its benefits. If a work includes AI-generated material, review the Office's guidance for works containing AI-generated content before filing.
Registration is not the same as technical protection, and it does not stop a model from collecting an image. It can be one part of a broader rights and enforcement strategy.
5. Use technical tools for the threat they address.
Glaze, developed by researchers at the University of Chicago, is designed to disrupt the imitation of an individual visual artist's style. It changes an image in ways intended to affect how models interpret style while keeping the work recognizable to people.
Glaze is not a universal shield. Its own documentation says it is aimed at style mimicry and is not consistently protective against image-to-image transformations. Effectiveness can vary by image, model, and future technical changes. Read the official Glaze FAQ before deciding whether it fits your work.
Use the tool for the threat it names, not the protection you wish it promised.
6. Publish deliberately.
Ask what a public post needs to accomplish. A portfolio preview may not need the same resolution or file access as a licensed delivery. Keep full-resolution masters offline when the public does not need them, and avoid posting source files that expose more than the finished work.
This can reduce easy direct reuse. It will not guarantee that a visible work is absent from a dataset, and it should not be described as one.
7. Monitor, document, then choose a response.
When you find a suspected copy or imitation, save the URL, date, screenshots, account details, and any surrounding description before contacting the platform or publisher. Keep your original files and public response separate from the evidence copy.
The right response depends on what happened. Platform reporting, a direct request, a licensing conversation, or advice from a qualified attorney may be appropriate. Do not let urgency erase the record you may need later.
A practical creator checklist.
- Original and draft files are backed up outside the publishing platform.
- Ownership, client rights, and AI-related permissions are written down.
- Your site or profile has a visible rights and contact statement.
- Current platform training and data-sharing settings have been reviewed.
- Any protection tool has been checked against its official limitations.
- You have a simple process for saving evidence of suspected misuse.
Sources and further reading.
- 01U.S. Copyright Office: Register Your Work
- 02U.S. Copyright Office: Registration guidance for works containing AI-generated content
- 03The Glaze Project: What Glaze is designed to do
- 04The Glaze Project: limitations and frequently asked questions