Can You Freely Use Text and Images Made with AI? Terms of Use, Copyright and Commercial Use
Whether you may use AI-made text and images is set by the service’s terms; whether they’re protected as your work depends on the creative expression a person added; and whether they infringe others’ rights depends on the actual output. Based on AI service terms, copyright standards and labeling rules, here is what to check before you post.

Can you freely use text and images made with AI? The question “May I use it?” should be split into three. Whether you may use it is in the service’s terms; whether it’s protected as your work depends on the creative expression a person added; and whether it infringes someone else’s rights depends on the actual output. Even when the terms allow use, you don’t automatically get copyright in the whole image, and if someone else’s work or character is in it, separate rights issues can arise. This post is based on public sources as of October 2026; rights in any individual output depend on how it’s used and what it contains.
Why split it into three questions?
Say you made a blog cover image with AI. If the terms allow it, you can use it within those conditions. That doesn’t mean you can stop others from using a similar image.
| Question | Where to check | Short answer |
|---|---|---|
| May I use it? | The terms of the service and plan you used | Yes, within the conditions the terms allow |
| Is it protected as my work? | The creative expression a person added | Using AI doesn’t rule everything out, but only human-added expression can be protected |
| Does it infringe others’ rights? | The actual output | The terms don’t guarantee it; check characters, logos, real people and similarity to existing works separately |

What do AI service terms say about output?
The terms of ChatGPT and Claude assign the company’s rights in outputs to the user to the extent the law allows. Some services and plans add conditions.
| Service | What the terms say (checked October 2026) |
|---|---|
| ChatGPT (OpenAI) | You own the output to the extent permitted by law; other users may receive similar output |
| Claude (Anthropic) | Assigns you its rights in outputs, “if any” |
| Midjourney | If you are a company, or work for one, with over US$1M a year in revenue, you need a Pro or Mega plan to own your assets |
So rather than relying on “AI images can be used commercially,” it’s more accurate to check the terms of the service and plan you actually used. One service’s terms can’t be applied to another. The difference between free and paid plans is covered in Free vs. Paid AI.
Does a little editing give you copyright?
Using AI doesn’t by itself remove a whole text or image from protection. What matters is what creative expression a person added.
- United States: In its January 2025 report, the U.S. Copyright Office distinguished output obtained from prompts alone from parts a person creatively modified or arranged.
- Korea: In June 2025 the Ministry of Culture, Sports and Tourism and the Korea Copyright Commission published a guide on registering works made with generative AI.
| Not enough on its own | Can be protected |
|---|---|
| Output obtained from a prompt alone | Text a person wrote (it can be considered separately from an AI image) |
| Picking the one result you liked | An original arrangement of several elements |
| Slightly changing size or color | Expression a person drew and added |
| Time spent or number of edits | The creative expression a person added, in that part |
In short, what creative expression you added matters more than how much you edited. Whether and how far it’s protected depends on the specific work.

How is being protected different from infringing others’ rights?
Even if your copyright in AI output is hard to establish, that doesn’t mean you can use the output anywhere freely.
| If | Check |
|---|---|
| It includes a famous character or another company’s logo | Copyright and trademark issues |
| It closely resembles a specific work | Whether it goes beyond mood to copy characters, composition and details. A similar art style alone doesn’t establish infringement |
| You uploaded a reference image | Whether you have the right to use that reference image |
| A real person’s face or voice | Portrait, personality and publicity rights. In Korea, since April 2022 the Unfair Competition Prevention Act bans commercial use of a well-known person’s name, face, voice and similar identifiers without permission |
| A company logo | Not just copyright, but whether it conflicts with other trademarks |
You can also describe the features you want, such as colors, materials and the feel of the lines, instead of naming an artist. And an AI saying “there’s no problem” doesn’t complete a rights check. AI answers themselves can be wrong; see Why Is AI So Confidently Wrong?.
Do you have to label something as AI-made?
The labeling duty in Korea’s AI Basic Act, in force since January 22, 2026, depends on who provides what. A user who makes an AI image for a blog is not in the same position as a business that provides a product or service with AI generation built in.
| Who | Rule |
|---|---|
| A business providing a product or service with AI generation | The AI Basic Act’s duty to label generative AI output |
| A user creating content with AI | Government guidance distinguishes ordinary content creation and use from providing AI services |
| Someone uploading to YouTube | Realistic generated or synthetic content and similar must be disclosed |
| Someone posting blogs or ads | Check the platform’s current policies |
So it’s a mistake to think everyone who used AI even a little must label it the same way. Still, if an image could be mistaken for a photo of a real person or event, saying it was made with AI helps readers judge. Labeling doesn’t resolve infringement issues.
What should you check before posting?
| Check | How |
|---|---|
| Blog images | Look at the terms, similarity of the output and labeling rules |
| Logos and product designs for sale | If it will represent your company, add a rights review suited to that use |
| Image search | Helps find similar works, but no results is not a guarantee |
| Work records | Keep originals, edits, materials used and your process. Records don’t create copyright, but they help explain what you made and changed |

In short
- Split AI output into three: may you use it, what expression did a person make, and are others’ rights involved.
- Check terms for the service and plan you used; one service’s terms don’t apply to another.
- Copyright depends on the creative expression a person added, not on edit count or time.
- Check characters, logos, real people and similarity to existing works separately from the terms.
- The AI Basic Act’s labeling duty distinguishes businesses providing services from users of output; check platform rules separately.
At ONE-PEX, building the AI-powered business system FROX, we believe it matters to decide in advance what to check before AI output is used in real work.
The images in this post were made with AI.
Sources (checked 2026-10-21)
- U.S. Copyright Office, Copyright and Artificial Intelligence (Part 2, 2025-01)
- Korea MCST press release on two generative AI copyright guides (2025-06-13)
- Korea Policy Briefing, AI Basic Act takes effect (2026-01)
- Shin & Kim newsletter on the AI transparency guidelines
- OpenAI, Terms of Use
- Anthropic, Consumer Terms of Service
- Midjourney, Terms of Service
- YouTube Help, Disclosing use of altered or synthetic content
Related reading: Why Is AI So Confidently Wrong? · Free vs. Paid AI · Where Can Your Personal Data Leak from ChatGPT?

A software planner who became a developer after two years of working with AI five to twelve hours a day. Today I build FROX, an AI-powered business system, at ONE-PEX.
More posts
- What Can You Build with AI?Orders in KakaoTalk, Stock in Excel, Receivables in a Notebook: Changing Messenger Orders Only the Owner Understands (Food Supplier, Part 1)
- What Can You Build with AI?I Fixed My TWA App, So Why Didn't It Change? What I Learned from the War with Caching
- What Can You Build with AI?Details Matter: Splitting 3D Characters and Vehicles So They Can Move (Building Road Voca 2, Part 5)
