Kind 1
What you wrote
Script, storyboard, narration, subtitles: the text you wrote yourself, and its versions.
Example: the third-draft script, and a note on what changed from the first two.
You delivered the film.Can you show what you created?
AI can help make the film, but it cannot explain which decisions, revisions, and choices came from you.
When a client, partner, or viewer asks, “What did you create in this film?”
PortraitPay helps you keep versions, key revisions, source information, and team roles as you create. So people can see not only the film, but what you contributed to it.
Records do not create copyright or guarantee client acceptance or legal outcomes.
When all you deliver is the finished film:
When all you have is the final video
People can see the result, but not your judgment, process, or contribution.
When a client needs answers
If you cannot quickly explain the process, sources, and team roles, a strong work may lose an opportunity.
When your work is questioned
When roles are unclear, the finished film alone may not show which creative decisions came from you.
You do not need to record every click. Keep these four kinds of material.
The example below is fictional. It illustrates the kinds of material and does not correspond to any real user, client, account, project, file path or asset.
Kind 2
Changes you made
The selections, rewrites, arrangements and editing decisions you made to generated output, and why you made them.
Example: the three shot versions that were used, and why the earlier one was replaced.
Kind 3
Where material came from
Reference material, music, fonts, and the models and tools you used, with the licence terms that attach to each.
Example: the licence scope of one music track, and the name and purpose of the generation tool.
Kind 4
Who took part, and what was delivered
The roles and attribution arrangements of the people involved, and which version actually went out to whom.
Example: the agreed attribution order, and which final cut was delivered.
What the finished film alone cannot say
The film alone
With only the finished film you can usually answer "what was delivered", but not "who exactly did what, and on what basis" — afterwards, memory is usually a vague impression.
The film plus process material
Adding process material adds four things: who wrote which version, who made which change, where the material came from, and which version was delivered. Those are specific facts a third party can check.
Record as you create, fix milestones, organize for delivery
This is a recommended practice, not a claim that every feature is connected, and no separate timestamp service is required per action.
Finish a step, capture a record
After a script revision, retain the adopted version, edit notes and sources; not every discarded generation.
Separate human work from AI use
Identify contributors, expressive edits, AI tools and uses, with before-and-after materials. A statement is not verification.
Fix key milestones
Separate claimed creation time, server recording time and external timestamp time. Label retrospective entries without backdating trusted time.
Organize records for their purpose
Organize selected versions, provenance and contributions; retain originals and verification data without treating private project records as public sharing content.
AI is already part of commercial video production
Clients, partners, and platforms increasingly care about: Who made this work, and which creative decisions came from you?
The industry research below illustrates this trend; the legal materials are presented separately in the U.S. rules section.
United States · Copyright Office report
Copyright and Artificial Intelligence, Part 2: Copyrightability
U.S. Copyright Office · Published 2025-01-29 · Source checked 2026-10-29
Nature: An agency policy report — not statutory law, and not a decision on rights in any individual case
The Report concludes that the outputs of generative AI can be protected by copyright only where a human author has determined sufficient expressive elements. That can include situations where a human-authored work is perceptible in an AI output, or a human makes creative arrangements or modifications of the output, but not the mere provision of prompts. It confirms that the use of AI to assist in the process of creation, or the inclusion of AI-generated material in a larger human-generated work, does not bar copyrightability, and that these questions can be resolved under existing law without legislative change. On prompts, the analysis states that given current generally available technology, prompts alone do not provide sufficient human control, while noting that highly detailed prompts could contain the user's desired expressive elements but at present do not control how the system processes them.
Scope: Applies to copyrightability analysis of AI output under United States law; the Report states these conclusions will be revisited as technology and law develop, and any particular work is still analyzed case by case
Our interpretation: Our interpretation: preserve specific writing, edits, selections and arrangements rather than generation counts.
Limits: This summarizes a US agency report, not a worldwide rule or a decision on any work. Contribution statements still need factual support.
Read original source — U.S. Copyright Office (new window)United States · Copyright Office announcement
NewsNet Issue 1060: Copyright Office Releases Part 2 of Artificial Intelligence Report
U.S. Copyright Office · Published 2025-01-29 · Source checked 2026-10-29
Nature: An official news release summarising the Report's conclusions
The release states that the outputs of generative AI can be protected by copyright only where a human author has determined sufficient expressive elements; that this can include situations where a human-authored work is perceptible in an AI output, or a human makes creative arrangements or modifications of the output, but not the mere provision of prompts; and that the use of AI to assist in the process of creation, or the inclusion of AI-generated material in a larger human-generated work, does not bar copyrightability. It also states that existing principles of copyright law are flexible enough to apply to this technology.
Scope: Applies to confirming the official wording and the publication date of Part 2
Our interpretation: Our interpretation: this official summary can be cited directly when explaining the United States position to a client.
Limits: The release summarises the Report; citing it implies no partnership or endorsement, and it does not change the Report's own limits.
Read original source — U.S. Copyright Office (new window)Industry survey · not law
AI is part of commercial video production
IAB · Published 2026-07-14 · Source checked 2026-10-29
Nature: An industry association survey report — not law or a regulatory document
IAB reports that nearly two in three surveyed buyers use GenAI for digital video creative, up from half in 2025.
Scope: Applies to showing how AI is used in commercial video production
Our interpretation: Our interpretation: documenting AI use and material sources belongs in commercial production records.
Limits: Source: IAB industry research. The data illustrates AI adoption in commercial video production and does not represent every creator or client.
Read original source — IAB (new window)Technical standard · not law
Timestamps bind a digest to a time
IETF / RFC Editor · Published 2001-08 · Source checked 2026-10-29
Nature: A technical standard — not law or a regulatory document
RFC 3161 specifies submitting a message imprint to a timestamp authority and receiving a signed token containing time information. It does not ask the authority to determine authorship.
Scope: Applies to how the timestamping mechanism itself works
Our interpretation: Our recommendation: record as you work and consider timestamping milestones. Individual or batch requests depend on cost, timing precision and service terms.
Limits: A timestamp is not copyright registration or proof that a statement is true. Later timestamping cannot prove an earlier claimed creation time. PortraitPay’s complete external timestamp workflow is not yet available.
Read original source — IETF / RFC Editor (new window)How U.S. copyright rules apply to AI-assisted work
This page covers copyright and record-keeping boundaries for the U.S. service flow only. Materials from other jurisdictions are not part of the current U.S.-only service and are not shown here.
United States
What the official material says
The Copyright Office concludes that whether AI output is copyrightable depends on whether a human determined sufficient expressive elements. Using AI to assist does not remove protection, and based on how currently available technology works, prompts alone do not provide sufficient control. That is not a final "prompts never suffice" rule: the Office states these conclusions will be revisited as technology and law develop, and whether a given human contribution amounts to authorship is analyzed case by case. Human expression perceptible in an output, the creative selection and arrangement of material in it, and creative modifications of it can all be protected.
Scope: copyrightability analysis under United States law. It is an agency report, not a statute, and not a decision on the rights in any particular work.
Copyright and Artificial Intelligence, Part 2: Copyrightability
U.S. Copyright Office · Published 2025-01-29 · Source checked 2026-10-29
Nature: An agency policy report — not statutory law, and not a decision on rights in any individual case
Scope: Applies to copyrightability analysis of AI output under United States law; the Report states these conclusions will be revisited as technology and law develop, and any particular work is still analyzed case by case
Read the original — U.S. Copyright Office (opens in a new window)NewsNet Issue 1060: Copyright Office Releases Part 2 of Artificial Intelligence Report
U.S. Copyright Office · Published 2025-01-29 · Source checked 2026-10-29
Nature: An official news release summarising the Report's conclusions
Scope: Applies to confirming the official wording and the publication date of Part 2
Read the original — U.S. Copyright Office (opens in a new window)
Show the longer explanation
Why keeping the before-and-after versions is more useful than counting generations: the Copyright Office turns on whether a human determined sufficient expressive elements, so the material that answers the question is what you wrote, what you changed, what you selected and how you arranged it — not how many times you called a model. Recording does not make a work copyrightable. What it does is give your explanation of your own contribution something to stand on.
PortraitPay does not provide legal advice. This is general information, not legal advice about any particular work, contract or dispute, and it does not replace the judgement of a qualified lawyer.
Current availability
The complete workspace flow for asset provenance, human contributions and version links is still being connected. External timestamp requests, receipt verification and controlled sharing must not be treated as available. A project snapshot is not a redacted public sharing page.
Arriving from reporting or comparison tools? Organize existing material without fabricating history. For future projects, record during creation. Supplementary analysis determines neither infringement nor takedown or dispute outcomes.
What recording buys you is lower organizing cost
Process material does not create copyright, cannot replace legal judgement, cannot prove that a statement is true, and does not decide how a dispute will end. What it does is make material easier to find, check and explain when you deliver, describe your rights or prepare for a dispute — because the facts were recorded when they happened instead of being recalled afterwards.
You can also do this without PortraitPay at all, organizing the same material in your own folders. The value of PortraitPay is only that it makes the work less effortful.
Start organizing project records
Use the workspace to keep the four kinds of material in the project as you go.
Start organizing project recordsLocal connector pilot (local-folder access only)
The invited-pilot allowlist and administrator switch apply only to the local connector, which needs a local process and folder authorization; they do not gate remote MCP. Remote MCP uses HTTPS and the host's browser OAuth flow to read cloud projects this account may access and records synced by the authorizing user. Check the host's MCP configuration docs for its capabilities.
See remote MCP and local connector instructions →PortraitPay does not provide legal advice. This is general information, not legal advice about any particular work, contract or dispute, and it does not replace the judgement of a qualified lawyer.