Download the Imageka app — create AI images on the go Get the Imageka app Get it on Google Play

Using your images

Using AI Images Commercially: Rights and Risks

What the model licence allows, why AI images may not be protected by copyright, and the trademark, likeness and disclosure rules to know before you publish.

By The IMAGEKA team Updated 13 min read

You can use images you make with IMAGEKA in commercial work, but "allowed to use" is not the same as "safe to use" or "yours to protect". This guide separates the three questions that matter before you publish: what the tool and model licences permit, what copyright law will and will not do for you, and which other laws (trademark, likeness, advertising, platform rules) can still catch you out.

General information, not legal advice

This page explains the general position in plain language as of October 2026. Laws differ between countries and are changing quickly, and courts are still deciding cases about AI. If an image matters to your business — a logo, packaging, a book cover, a paid campaign — check with a qualified lawyer in your jurisdiction before relying on it.

Key takeaways
  • IMAGEKA claims no ownership of your outputs and lets you use them commercially, subject to our Terms and the law.
  • Every model IMAGEKA uses is an open-source model whose licence permits commercial use. Some of those licences add use-based restrictions (no illegal or harmful uses), which apply to you through our Terms.
  • Permission to use is not protection. In the US and many other places, a purely AI-generated image is unlikely to be protected by copyright, so you may not be able to stop others copying it.
  • Human creative work — substantial editing, compositing, or selecting and arranging images into a larger work — can be protectable, even when the raw outputs are not.
  • Trademark, likeness, consumer-protection and platform rules apply regardless of how an image was made. Most real-world problems come from these, not from copyright.

What IMAGEKA's terms allow

Our position is simple: we do not claim ownership of the images you generate, and you may use them for personal or commercial purposes — on a website, in an advert, on a product, in client work — without paying us or crediting us. That permission is subject to our Terms of Service and to the law, which matters in two ways.

First, the Terms prohibit certain uses outright, such as content that infringes someone else's rights, impersonates real people, or breaks the law. A licence from us cannot authorise something the law forbids.

Second, we can only give you what we have. We can promise not to assert any claim over your images; we cannot promise that anyone has copyright in them, or that an output does not accidentally resemble something protected. Those are questions of law and of what the image actually shows, and they are covered below.

Two practical points that are easy to miss:

  • Keep your own copies. IMAGEKA stores generated images only in your browser's local storage. If you clear your browser data, they are gone, and we cannot recover them. Download anything you intend to use.
  • Keep a record. For anything commercial, save the prompt, the date and the file alongside your project. If a platform, client or publisher later asks how an image was made, you can answer accurately.

The model licences: open models that allow commercial use

IMAGEKA does not rely on a single model. Each request is routed automatically to one of several open-source (open-weight) image models, chosen by your style preset and wording: a fast general-purpose model, or a specialist for photorealism, anime/illustration or painterly styles. Every model we use is released under a licence that permits commercial use. These are permissive licences of the kind used across open image models, such as Apache 2.0 and CreativeML OpenRAIL++-M.

The two types work a little differently. Apache 2.0 is a general open-source licence that allows commercial use, modification and redistribution. OpenRAIL-style licences also allow commercial use, but add use-based restrictions: the model and its outputs must not be used for illegal purposes, to harm or exploit people, to generate defamatory content or harmful misinformation, and similar. Those restrictions pass through to you via our Terms of Service, so they apply whichever model made your image. In practice they line up with what the Terms and the law already prohibit; ordinary commercial work such as adverts, product listings, book covers or client projects is not affected.

It is worth being precise about what these licences cover. They govern the model software and weights. They do not grant you rights in any particular output, and they do not promise that an output is free of third-party rights. They simply mean there is no licence restriction from the model makers standing between you and lawful commercial use.

If you want to understand how these models turn a prompt into an image, see How AI image generators work.

This is the part most "commercial licence" marketing leaves out. Being allowed to use an image is one thing. Being able to stop other people using it is another, and that second right normally comes from copyright. For images generated mainly by AI, copyright protection is limited or absent in many jurisdictions.

The United States

US copyright law protects works of human authorship. Three developments set out where things stand:

  • Copyright Office guidance. In January 2025 the US Copyright Office published Part 2 of its report on copyright and AI, dealing with copyrightability. Its conclusion is that prompts alone, as the technology currently works, do not give the user enough control over the expressive result to make them the author of the output. Typing "a red fox in a snowy forest at dawn" and accepting what comes back does not make you the author of that picture in the Office's view.
  • What can be protected. The same report confirms that human contributions can be protected: your own artwork used as an input where it remains perceptible in the output, creative selection and arrangement of AI material, and modifications substantial enough to be original expression. The protection covers the human contribution, not the AI-generated parts on their own.
  • The courts. In Thaler v. Perlmutter, the US Court of Appeals for the D.C. Circuit held in March 2025 that the Copyright Act requires a human author and upheld the refusal to register an image listed as created autonomously by an AI system. The Supreme Court declined to hear an appeal in March 2026. That case concerned a work claimed to have no human involvement at all, so it does not settle exactly how much human input is enough — but it confirms the baseline.

When registering a work that includes AI-generated material, the Copyright Office expects applicants to disclose it and to claim only the human-authored parts.

Elsewhere

Other countries take different approaches. Most copyright systems, including in the EU, tie protection to an author's own intellectual creation, which points the same way as the US. The UK has an unusual provision for "computer-generated works" with no human author, but how it applies to modern generative AI is uncertain and has been under government review. Courts in some countries have granted protection where a user made detailed, iterative creative choices. If you sell internationally, do not assume one country's answer applies everywhere.

What this means in practice

  • If you publish a raw, unedited output, competitors may be able to reuse it, and you may have little legal recourse.
  • For a logo or brand mark, a raw AI image is a weak foundation. Use it as a sketch, then have a designer redraw it, and protect the result as a registered trademark (which depends on use in trade, not on copyright).
  • For products where exclusivity is the value — art prints, illustrated books, merchandise designs — meaningful human work on top of the output is what gives you something you can defend.
  • Expect others to generate similar images. The same model with a similar prompt can produce close variations. Distinctiveness comes from your editing, composition and context.
More human creative input, more likely to be protectable Raw output prompt only, used as-is Light edits crop, colour, small retouching Selection & arrangement layout, collage, book or comic Substantial work repainting, compositing Protection, where it exists, covers the human contribution — not the AI-generated material itself.
A simplified view of the US approach. Where the line falls in a specific case is decided case by case.

Trademarks, characters and other people's work

An image you are allowed to use can still infringe someone else's rights because of what it shows. The tool that made it is irrelevant to that question.

  • Logos and brand names. A generated scene with a recognisable sports brand on a shoe, a famous soft-drink logo on a can, or a car maker's badge can create trademark problems, especially in advertising where it implies endorsement. Remove or replace them.
  • Characters. Well-known cartoon, film and game characters are protected by copyright and often by trademark. Prompting for a character by name, or describing one so precisely that the result is recognisable, gives you an image you should not sell or use in marketing.
  • Artists' work. Artistic style as such is generally not protected, but an output that closely reproduces a specific existing artwork, photograph or design can be infringing. We recommend describing the visual qualities you want (medium, palette, lighting, composition) rather than naming living artists — it is better practice and usually produces a more controllable result. Our prompt-writing guide shows how.
  • Accidental resemblance. Generators sometimes produce shapes that look like real logos, or garbled text that resembles a brand name. Zoom in on signage, clothing, packaging and backgrounds before you publish.

Real people, likeness and personality rights

Many jurisdictions protect a person's name, image and likeness against commercial use without consent. In the US this is mainly state "right of publicity" law, and several states have added specific rules about digital replicas and deepfakes. In Europe, data-protection law and national personality or image rights apply. Advertising standards bodies in many countries also prohibit implying that a real person endorses a product when they do not.

The practical rules:

  • Do not generate images of identifiable real people — celebrities, politicians, your competitor's CEO, or a private individual — for commercial use. IMAGEKA's Terms prohibit impersonating real people.
  • Be aware that a generic "person" can, by chance, resemble a real individual. For a single ad featuring one prominent face, a licensed stock photo or a photo shoot with a signed model release may be the safer choice.
  • Never create images that place a real person in a sexual, defamatory or misleading situation. Beyond civil liability, this can be a criminal offence in a growing number of countries.

Misleading advertising and consumer protection

Consumer-protection law in most countries prohibits advertising that is likely to mislead, whether it uses a photograph, an illustration or an AI image. Regulators have been clear that "AI made it" is not a defence. Common traps for small businesses:

  • Products you do not sell. A generated image of a dish, room, garment or gadget that looks better than, or different from, what the customer will actually receive. If you sell it, photograph it. Use AI for backgrounds, mood and concepts, not as a substitute for showing the real product.
  • Fake testimonials and customers. Pairing an invented face with a review, or showing "happy customers" who do not exist as if they were real, is the kind of practice regulators treat as deceptive. In the US, the Federal Trade Commission's rule on fake reviews and testimonials, in force since late 2024, targets exactly this.
  • Before-and-after images. Generated or AI-enhanced "results" for cosmetics, fitness, cleaning products or renovations imply a real outcome. Do not use them.
  • Locations and events. A generated image of your hotel's "view" or a "packed" event that never happened misrepresents what you offer.
Tip

A useful test: if a customer later learned how the image was made, would they feel misled about what they bought? If yes, use a real photo or label the image clearly as an illustration. Our small-business recipes focus on uses that pass this test, such as backdrops, blog headers and event artwork.

Platform and marketplace rules

Even where the law is silent, the platforms you sell or publish through have their own rules, and they can remove listings or close accounts. Policies change often, so treat this as a map of what to look for rather than a summary of current terms:

  • Stock photo sites. Some accept AI-generated submissions only if they are labelled as such and meet extra conditions (no real people or brands, sometimes a requirement that you hold rights in the prompt and inputs); others do not accept them at all.
  • Print-on-demand and handmade marketplaces. Many require you to disclose AI involvement in the listing, and marketplaces built around handmade goods may restrict how AI-generated designs can be described or sold. Designs that use characters or brands are removed regardless.
  • Book publishing. Major self-publishing platforms commonly ask whether a cover or interior includes AI-generated content at upload. Traditional publishers and literary competitions set their own rules, sometimes excluding AI artwork entirely.
  • Advertising networks and social platforms. Several require a label on realistic AI-generated or altered content, particularly in political or social-issue ads, and some apply labels automatically when they detect metadata.

Read the current policy of each platform before you upload, and answer disclosure questions truthfully — inaccurate answers are a common reason for account suspensions. For sizing and format advice per platform, see AI images for social media.

Disclosure and labelling trends

The direction of travel is towards more transparency about synthetic media, through both law and technical standards.

The EU AI Act (Article 50)

Article 50 of the EU AI Act sets transparency obligations that apply from 2 August 2026. In outline:

  • Providers of generative AI systems must ensure outputs are marked in a machine-readable way and detectable as artificially generated. Under the "digital omnibus" amendments politically agreed in May 2026, systems already on the market before August 2026 were given a short grace period for this marking duty, reported as running to 2 December 2026.
  • Deployers — businesses and professionals who use AI to publish content — must disclose when an image, video or audio is a "deepfake", meaning it resembles real people, places, objects or events and would falsely appear authentic. There are lighter requirements for evidently artistic, satirical or fictional works.

The European Commission has also published a voluntary code of practice and optional icons for labelling AI-generated content. If you publish to an EU audience, a realistic image of a "real" scene is where disclosure matters most. A clearly illustrative image (a watercolour of a teapot) is a different case from a photorealistic image of an event that never happened.

Content Credentials (C2PA)

The C2PA standard, promoted as "Content Credentials", attaches signed metadata to a file recording how it was created and edited. Several major generators, cameras and editing apps now add it, and some social platforms read it to apply "AI-generated" labels. Separately, some providers embed invisible watermarks. Be aware that metadata is often stripped when images are screenshotted, re-saved or uploaded to some services, so it is a helpful signal rather than a guarantee — and that removing provenance information deliberately to pass off an image as authentic may breach platform rules.

Labelling voluntarily is usually low-cost. A short note such as "Illustration created with AI" in a caption or image credit is enough in most contexts and protects trust if a customer asks.

A pre-publication checklist

Run through this before any AI image goes into paid, printed or client-facing use:

  1. Inspect at full size. Look for logos, readable brand names, distorted text, extra fingers and odd artefacts. The troubleshooting guide covers fixes.
  2. No real people or characters. Does any face look like someone identifiable? Does anything resemble a known character, mascot or artwork?
  3. Accurate representation. Does the image show something you actually sell, offer or did? If not, is it obviously illustrative?
  4. Exclusivity needed? If you must stop others copying it (logo, product design, book cover), plan meaningful human work on top, and consider professional advice.
  5. Platform rules checked. Have you read the current AI policy for every place you will publish or sell, and answered disclosure questions honestly?
  6. Disclosure decided. Is the image realistic enough that viewers could think it is a real photo of a real event? If so, label it.
  7. Records kept. Saved the file, prompt and date? Remember IMAGEKA keeps images only in your browser.
  8. Client informed. If you are delivering work to a client, tell them which elements are AI-generated and what that means for their ability to protect them. Many contracts promise "original work" and full assignment of rights — check yours does not promise more than you can give.

Frequently asked questions

Do I own the images I make with IMAGEKA?

We claim no ownership and let you use them commercially under our Terms. Whether anyone owns copyright in a purely AI-generated image is a separate legal question; in the US and many other countries the answer for raw outputs is likely no. Your own creative edits and arrangements can be protected.

Can I sell prints or merchandise with IMAGEKA images?

Our Terms allow it, and the licences of the models we use permit commercial use. You still need to avoid characters, brands and real people, follow the marketplace's AI-disclosure rules, and accept that others may be able to copy an unedited design.

Do I have to credit IMAGEKA or the model makers?

No attribution to IMAGEKA or to the makers of the underlying models is required for your images. Separately from credit, some platforms and laws may require you to disclose that content is AI-generated; that is about transparency to viewers, not attribution to us.

Can I use an IMAGEKA image as my logo?

You can use it as a starting point, but a raw output makes a weak logo: it may not be protected by copyright, someone else could generate something similar, and it may be hard to reproduce cleanly at small sizes. Have a designer redraw the concept and register the final mark as a trademark if it matters to your business.

Can I register copyright in an AI image?

In the US you can register a work that contains AI material, but you must disclose the AI-generated parts and claim only your human contribution, such as substantial modifications or the selection and arrangement of elements. The AI-generated material itself will not be covered.

Is it legal to use AI images in adverts?

Generally yes, as long as the advert is not misleading, does not use someone's likeness without consent, and does not infringe trademarks. Do not use AI images to depict products, results or testimonials that are not real, and check any platform-specific labelling rules for realistic synthetic content.

Does the EU AI Act mean I must label every AI image?

Not every image. The deployer duty in Article 50 focuses on deepfakes — realistic content that could falsely appear authentic — with lighter rules for clearly artistic or fictional work. Platform rules may go further. Labelling realistic images voluntarily is a sensible default.

Could an image I generate infringe someone's copyright?

It is possible, especially if your prompt names a specific character, artwork or brand, or the output happens to closely resemble an existing work. Describe what you want in your own terms, inspect results carefully, and do not publish anything that looks like a recognisable existing work. See the glossary for the technical terms used in these discussions.