When Midjourney images move from experimentation into client work, campaigns, products or brand materials, the practical challenge is not simply “protect the image.” The more useful goal is to keep a clear record of what was created, what rights apply, what a person changed, where the asset was used and who approved the final version.
This operational checklist complements our broader Midjourney copyright and commercial-use guide. It is not legal advice and does not promise that recordkeeping, watermarks or editing automatically create copyright protection.
1. Record the project purpose before generating assets
Start by classifying the intended use. A temporary blog illustration, an internal concept image, an advertising creative, merchandise art and a permanent logo have different commercial consequences.
Record:
- project or client name;
- intended channels;
- whether the asset needs to be exclusive;
- whether it will be sold, licensed or transferred;
- whether recognizable people, brands or protected characters are prohibited;
- who has final approval.
This prevents an experimental image from quietly becoming a high-value brand asset without a rights review.
2. Save the Midjourney terms and plan information that apply
Platform terms can change. Before a significant commercial release, check the current Midjourney Terms of Service and note the account plan used for the project.
For important work, retain a PDF or dated internal record of the relevant terms rather than relying on memory months later.
3. Keep the original generation history
Preserve enough material to reconstruct the creative process:
- selected prompts where relevant;
- source generations;
- variations and upscales used;
- generation dates;
- the account or workspace used;
- downloaded original files.
These records do not automatically establish copyright, but they are useful project documentation and help distinguish source output from later human work.
4. Document human-authored contributions separately
Under current U.S. Copyright Office guidance, copyrightability turns on human authorship. Prompts alone generally do not give a person sufficient control over expressive elements to make the generated output copyrightable. Human-authored material, creative arrangement and qualifying modifications can be relevant depending on the work.
For primary guidance, see the U.S. Copyright Office AI study.
Keep editable files that show meaningful human work, such as:
- original drawings or photography incorporated into the composition;
- layered Photoshop, Affinity or other editing files;
- custom typography and layout;
- manual compositing of several elements;
- substantial retouching or reconstruction;
- creative selection and arrangement in a larger human-authored work.
Do not use a made-up percentage such as “20% edited” or “30% human” as a legal threshold. Copyrightability is not determined by a universal editing percentage.
5. Run a third-party-rights review before publication
Look at the final asset, not just the prompt. Check for:
- recognizable trademarks or logos;
- well-known fictional characters;
- celebrity or private-person likenesses;
- distinctive product designs or trade dress;
- third-party photographs or artwork you intentionally introduced;
- text that may contain names, slogans or inaccurate claims.
If the image is commercially important and a potential conflict is unclear, obtain qualified advice rather than assuming the AI platform has cleared the rights.
6. Treat logos and core identity assets as a special case
If an asset will function as a logo, mascot or central product identity, the business may need durable exclusivity and trademark clearance. A raw AI generation is often a poor place to assume those rights will automatically exist.
For core identity work, consider a human-led design process with documented authorship and a trademark review appropriate to the markets where the brand will operate.
7. Use watermarks for distribution control—not as a copyright shortcut
A watermark can discourage casual reuse, identify a preview or distinguish an approval copy from a final file. It does not create copyright where copyright otherwise does not exist.
Useful watermark practices include:
- watermarking client previews rather than final licensed files;
- using lower-resolution approval copies where appropriate;
- keeping clean masters in controlled storage;
- recording who received unwatermarked files.
8. Separate working files from delivery files
Use a predictable folder structure so the final asset can be traced back to its source and approvals. For example:
01-brief02-midjourney-source03-human-edits04-review05-final06-license-and-terms
This is ordinary production hygiene, but it becomes especially valuable when a project combines generated and human-authored material.
9. Put client rights and responsibilities in the contract
When work is created for a client, the contract should address what is being delivered and what rights are being transferred or licensed. Avoid promising exclusive copyright in material when you do not know whether that exclusivity exists.
Depending on the engagement, a contract may need to explain:
- whether generative AI may be used;
- which party approves that use;
- what editable/source files are delivered;
- what human-created elements are included;
- how third-party materials are handled;
- who is responsible for final legal clearance;
- whether exclusivity is promised and on what basis.
Contract language should be reviewed for the actual jurisdiction and transaction when the stakes justify it.
10. Keep approvals that show the final version
Email threads and chat messages can be ambiguous when many image versions exist. For commercial projects, record the exact final filename or asset ID approved for release.
Useful approval information includes:
- final filename and version;
- approval date;
- approving person;
- approved channels or uses;
- any restrictions or required attribution.
11. Do not overstate what “ownership” means
Midjourney’s current terms discuss ownership of assets to the fullest extent possible under applicable law. That contractual language does not replace the legal requirements for copyright protection, nor does it remove third-party rights.
Use precise internal language. “We may use this asset under the current service terms” is a different statement from “we own an enforceable copyright in every element of this image.”
12. Keep a release log for important assets
For a high-value project, maintain a simple release record showing where the asset appears:
- website URL;
- campaign or ad account;
- social channel;
- print or merchandise use;
- client delivery;
- date first published;
- date retired, if applicable.
This makes future replacement, takedown or rights review substantially easier.
13. Review the asset again if its use changes
An image approved as a blog illustration should not automatically become packaging, merchandise or a logo. Re-run the rights review if the commercial importance or distribution changes.
14. Use reverse-image search as a monitoring tool, not a legal conclusion
Reverse-image search can help find public copies or visually similar images. It cannot by itself tell you whether a use is lawful, whether the other party created a similar image independently or whether you have an enforceable claim.
If you find a concerning use, preserve evidence before contacting anyone and obtain appropriate advice when the commercial stakes are meaningful.
15. Maintain a simple asset-rights record
A spreadsheet or project database can be enough. Useful fields include:
| Field | What to record |
|---|---|
| Asset ID | Unique internal filename or identifier |
| Purpose | Blog, ad, product, client deliverable, etc. |
| Generation date | When the Midjourney source was created |
| Plan/terms checked | Date and plan used |
| Human contributions | Summary and link to editable files |
| Third-party review | Brands, likenesses, licensed inputs checked |
| Approval | Who approved the final asset and when |
| Published uses | Where the asset was released |
What this checklist cannot do
Good records reduce confusion and support a professional workflow. They do not guarantee that an asset is copyrightable, non-infringing or exclusive. They also do not replace legal review for a disputed or commercially significant use.
Bottom line
For commercial Midjourney work, the most useful protection is disciplined asset management: know the current platform terms, document human-authored work, check third-party rights, keep approvals and contracts clear, and revisit the review when an asset moves into a higher-risk use.
For the legal concepts behind this workflow, read Midjourney Copyright and Commercial Use: What Creators Should Check.