Who owns content made with AI? In many countries copyright protects work created by a human author, so material generated almost entirely by AI may not be protected by copyright at all, even if the tool’s terms say you “own” the output. The more genuine human creative choices you add, such as selecting, arranging and editing, the stronger your position, but the rules are still being worked out.
This article explains the main ideas in plain language for designers, marketers and small business owners. It is not legal advice: for contracts, disputes or high-value work, talk to a lawyer who knows intellectual property law in your country.
Why is human authorship the key idea in AI copyright?
Copyright systems were designed around people. They reward a person’s creative choices: the words chosen, the composition of a photograph, the arrangement of a song. Most legal systems, Vietnam’s included, describe the author as the person who directly creates the work. AI tools do not fit neatly into that picture.
In the United States, the copyright office has taken the position that material generated by AI without sufficient human control cannot be registered, while the human contributions in the same work can be. Courts there have also rejected attempts to list an AI system as the author. Other countries are at different stages, and some have older provisions on computer-generated works that are now being reinterpreted.
What counts as sufficient human contribution is the hard part. Typing a one-line prompt is generally treated as supplying an idea rather than making the expression. Choosing among many outputs, combining them, painting over parts, heavily rewriting text, or using AI for one element inside a larger human-made work all add human authorship, but there is no precise threshold.
What the human-authorship gap means for businesses
This creates a practical gap for businesses. If a logo or a slogan was produced almost entirely by AI, you may struggle to stop a competitor from copying it. That is a strong reason to have a human designer develop key brand assets, using AI only to explore directions.
Because these positions are changing, treat any summary, including this one, as a starting point. Check the current guidance of the copyright authority in each country where you need protection.
What do AI tool terms of service actually decide?
Ownership clauses. Many AI providers state that, between you and them, you own the output. That is a contractual promise from the provider; it cannot create copyright where the law does not recognise any.
Commercial use permissions. Free plans sometimes restrict commercial use or require attribution. Read the current terms for the plan you actually use, because they change.
Licences back to the provider. Some terms give the provider rights to use your inputs and outputs, for example to improve the service. Check this before uploading client materials.
Indemnity offers. Certain business plans promise to defend customers against copyright claims over outputs, usually with conditions such as keeping safety filters on. Know the conditions before relying on the promise.
Content restrictions. Terms usually forbid generating material that infringes others’ rights or impersonates real people. Breaking them can cost you access to the tool on top of any legal risk.
Non-exclusive outputs. Another user may receive a very similar result from a similar prompt. You cannot assume uniqueness, which matters a great deal for logos and brand assets.
Platform rules. Stock photo sites, print-on-demand shops and app stores have their own policies on AI content, often requiring disclosure. Follow the platform’s rule as well as the tool’s.
Jurisdiction and updates. Terms state which country’s law applies and can change with notice. Save a copy of the terms in force when you create important work.
The training-data dispute and the risks in AI output
Ownership is only half the question. The other half is whether the output infringes someone else’s rights.
| Risk | What it means | Practical habit |
|---|---|---|
| Output resembles an existing work | A generated image or passage can come close to a specific protected work, especially famous images or when prompted with an artist’s name. | Avoid prompting with living artists’ names or famous titles; reverse-search key visuals before publishing. |
| Characters and trademarks | Tools can reproduce well-known characters, logos or product designs that belong to someone else. | Never use recognisable characters or brands in commercial material without a licence. |
| Real people’s likeness | Faces and voices of real people raise personality and privacy issues separate from copyright. | Get written consent before using anyone’s face or voice, including your own staff. |
| Training-data lawsuits | Authors, artists and publishers in several countries have sued AI companies over using their works for training; many cases are unresolved. | Follow developments; for sensitive commercial uses, prefer tools that disclose licensed training data. |
| Text that copies sources | Generated text may repeat phrases from sources or state facts without attribution. | Check key passages with a plagiarism tool and add sources yourself. |
| Code licences | Coding assistants can suggest code resembling open-source code released under licences with conditions. | Turn on duplicate-code filters where available and review licences for larger snippets. |
None of these risks makes AI output unusable. They mean the person who publishes remains responsible for what is published.
Habits that protect creators and small businesses
Keep a simple record of how important work was made: the prompts, the outputs you picked and the edits you made afterwards. If ownership is ever questioned, a record of human choices is your best evidence.
Use AI as a starting point for brand-critical assets, then let a human develop the final version. For a logo, that might mean using AI to explore directions and asking a designer to draw the chosen concept from scratch.
Be transparent with clients. State in your quote or contract whether AI tools are used, who owns the result and what you guarantee. Clients increasingly ask, and surprises damage trust more than the tool itself does.
Respect other creators even where the law is unclear. Not imitating a named living artist’s style for commercial work, crediting sources and paying for licensed material are good practice however the court cases end.
When money or reputation is significant, such as a large campaign, a publishing deal or a dispute over copying, consult an intellectual property lawyer. A short consultation is cheap compared with rebranding after a conflict.
Frequently asked questions
Can I copyright an image I made with AI?
It depends on the country and on how much human creative contribution the work contains. Purely AI-generated images may not qualify where a human author is required, while your own edits and arrangement may be protectable.
If the AI company says I own the output, am I safe?
That clause means the provider will not claim it, but it cannot guarantee legal protection or that the output does not infringe someone else’s work. You remain responsible for what you publish.
Is it legal to use AI to imitate a famous artist’s style?
Style alone is generally not protected, but outputs that copy specific works can infringe, and imitating a living artist for commercial gain raises ethical and reputational problems. Keep artists’ names out of prompts for commercial work.
Do I need to disclose that content was made with AI?
Some platforms and some laws already require labels, especially for realistic images and for advertising or political content, and these rules are expanding. Even where it is not required, disclosure builds trust with clients and audiences.
Who is liable if AI output infringes copyright?
Usually the person or company that publishes or sells it, although contracts and indemnity terms can shift some of the cost. For a real dispute, consult an intellectual property lawyer.
