Dough
All articles
Resource library

Founder guide

Who owns a product created with AI?

How to separate brand, design, invention, supplier, and platform rights when AI contributes to a physical product, and what Dough’s terms say about it.

Updated 2026-08-07Founders using AI to develop product concepts, packaging, or marketing
Packaged physical products moving through a production line

Direct answer

There is no universal answer that depends only on whether AI was used. Ownership comes from the platform terms, the human contribution, employment and contractor agreements, supplier contracts, existing third-party rights, and the law where you operate. On Dough specifically, the terms state that Dough does not claim ownership of the product formulations, recipes, or specifications developed for you.

Separate the rights involved

One product can carry several different rights at once, and they do not travel together. Trademarks live in the brand, copyright in artwork and copy, design rights or design patents in appearance, utility patents in functional inventions, trade secrets in formulas and processes, and contractual rights in supplier files and tooling.

Most confusing ownership arguments are really an argument about which of these is being discussed. Name the specific right before asking who owns it.

Read every agreement in the chain

Check the AI provider terms, the platform terms, employee and contractor invention assignments, design-service agreements, supplier terms, tooling ownership, and the licenses covering fonts, photographs, software, and any source material you fed in.

The weakest link governs. A generous platform license does not help if the freelancer who drew your label never assigned their copyright to you.

What Dough’s terms say

Dough’s platform terms state that intellectual property rights in products developed through the platform are governed by the agreement between you and the development partner, and that Dough does not claim ownership of product formulations, recipes, or specifications developed for you. There is a stated carve-out for aggregated and derived data, which the terms describe separately.

That is the part founders most often need to hear plainly, and it is also the part you should verify yourself rather than take from a marketing page. The terms are linked below.

Read the current terms directly. This page summarizes them and can lag behind changes.

AI output can still create infringement risk

A right to use output under one contract does not guarantee that the output avoids somebody else’s trademark, copyright, patent, publicity, or design rights. These are different questions with different answers.

Run clearance appropriate to the category before you invest in packaging or production. A name is cheap to change before the print run and very expensive afterwards.

Keep an evidence trail

Preserve the original brief, your edits, the drafts you rejected and the one you selected, designer files, supplier drawings, approvals, and dates. That record is what establishes provenance and human contribution later.

A workflow that keeps drafts, the selected version, price history, and manufacturing approvals in one account gives you most of that trail as a side effect of working. Reconstructing it from screenshots and chat logs a year later does not go well.

Get jurisdiction-specific advice

Rules on AI authorship and inventorship are still developing, and they differ by country. A qualified lawyer should review any material trademark, patent, regulatory, or ownership question before you launch.

The cost of that review is small next to a rebrand after a cease-and-desist, or a production run you cannot legally sell.

This guide provides general information, not legal advice.

Sources and product references