You found an image that fits your page perfectly. Now you have to work out whether you are allowed to use it, and the page you found it on says something like "royalty-free" or "CC BY-NC-SA 4.0" or nothing at all. This post explains what those labels actually mean in plain language, so you can tell the difference between an image you can publish, an image you can publish with credit, and an image that will get you a letter from a licensing agency.
The default is that you cannot use it
Copyright is automatic. The moment someone takes a photo, draws an illustration, or renders a graphic, they own the copyright in most countries, with no registration, no notice, and no copyright symbol required. Registration matters for what you can claim in court in some places, notably the United States, but it is not what creates the right.
So the correct default assumption for any image is: someone owns this, and I do not have permission. Every license category below is an exception to that default, granted deliberately by the owner. If you cannot identify which exception applies, you do not have one.
That sounds strict. It is also why so many small sites get an invoice years later for a photo somebody grabbed off a search results page in 2019.
All rights reserved, and what "contact for permission" really means
"All rights reserved" is the plainest label: the owner has not granted anyone a general license. To use the image you negotiate directly. That negotiation produces a license, which is simply a written permission with conditions attached.
A few things people get wrong here:
- Asking and getting no reply is not permission. Silence is not a yes.
- Permission from the wrong person is not permission. The subject of a photo usually does not own it. The photographer does, unless it was work made for hire or the rights were assigned in writing.
- Permission for one use is not permission for all uses. If a photographer let you use a shot in a newsletter, that does not automatically cover a billboard.
Get permission in writing, and get it specific: what image, what uses, what territory, how long, and whether you can crop or edit it.
The Creative Commons family, condition by condition
Creative Commons licenses are standard, pre-written permissions that any creator can apply to their own work. There are six of them, built by combining four conditions. The current version is 4.0, published in 2013, though you will still find 2.0 and 3.0 material in the wild.
The four conditions:
- BY (Attribution). You must credit the creator. Every current CC license includes this.
- SA (ShareAlike). If you adapt the work, you must release your adaptation under the same license. This is the condition that surprises people: it can pull your own derived work into the same terms.
- NC (NonCommercial). You may not use it primarily for commercial advantage or monetary compensation.
- ND (NoDerivatives). You may share the work, but you may not publish a modified version.
Combine them and you get the six licenses:
| License | Credit required | Commercial use | Edits allowed | Adaptation must match license |
|---|---|---|---|---|
| CC BY | Yes | Yes | Yes | No |
| CC BY-SA | Yes | Yes | Yes | Yes |
| CC BY-ND | Yes | Yes | No | Not applicable |
| CC BY-NC | Yes | No | Yes | No |
| CC BY-NC-SA | Yes | No | Yes | Yes |
| CC BY-NC-ND | Yes | No | No | Not applicable |
Two of these conditions are genuinely fuzzy at the edges, and pretending otherwise does you no favors.
NonCommercial is defined in the license text as use not primarily intended for or directed toward commercial advantage or monetary compensation. That is clear for a school project and clear for a magazine advert. It is much less clear for a personal blog that runs display ads, a nonprofit's fundraising page, or a free app with a paid tier. Creative Commons itself acknowledges the line is contextual. If your use sits in that grey zone, either get separate written permission or pick a differently licensed image.
NoDerivatives raises the question of what counts as a derivative. Under the CC licenses, a change is an adaptation when it would produce a copyrightable derivative work under the applicable law. Purely technical changes, converting a PNG to WebP or resizing it to fit your layout, are generally not adaptations. Cropping to change the composition, recoloring, adding text over it, or combining it into a collage is much more likely to be. If a license says ND and you want to crop for meaning rather than for format, ask.
Public domain and CC0 are related but not identical
Public domain means no copyright applies, so there is nothing to license. Works land there in three main ways: the term expired, the work was never eligible for copyright, or the owner dedicated it to the public.
Term rules differ by country and are more complicated than any single sentence can capture. Two broad markers: in much of the world, copyright in a work by a known author lasts for the author's life plus 70 years. In the United States, works published before a rolling 95-year cutoff are in the public domain, which in 2026 means works published in 1930 or earlier. Works of the US federal government are generally not protected by copyright at all.
CC0 is not a license in the ordinary sense. It is a dedication in which the creator waives their copyright as far as the law allows, and where a full waiver is not possible, grants a license as permissive as the law allows. Practically, treat CC0 material as usable without attribution and without restriction from the copyright holder.
Two caveats that trip people up:
- CC0 covers copyright. It does not waive trademark or patent rights, and it cannot waive rights held by third parties, such as the personality rights of a person shown in the photo.
- A photograph of a public domain painting may still carry its own copyright claim in some jurisdictions, depending on how much originality the photography involved. This is genuinely disputed and has been litigated differently in different countries.
Stock sites with their own custom terms (the Unsplash, Pexels, and Pixabay licenses, for example) are permissive but not identical to CC0. They typically ban reselling the image as-is, using it in a competing stock library, and using it to imply endorsement by an identifiable person. Read the specific terms once. They are usually short.
"Royalty-free" means one fee, not no fee
This is the single most misunderstood term in the whole subject. Royalty-free describes the pricing model, not the price. You pay once, then you can use the image many times without paying a further royalty for each use.
You still pay. And you still have terms. A typical royalty-free stock license restricts:
- Redistribution. You cannot sell or give away the image file itself, only the work you made with it.
- Print runs or impressions above a stated ceiling, above which you need an extended license.
- Products for resale. Putting the image on mugs, posters, phone cases, or a website template you sell almost always requires an extended or merchandise license.
- Sensitive-use contexts. Many licenses forbid using an image of an identifiable person in connection with topics like health conditions, adult content, or crime, without an additional disclaimer or release.
- Transferability. The license usually belongs to the buying entity. If an agency bought it for one client, it may not cover a second client.
Royalty-free is also almost always non-exclusive. Your competitor can buy the same photo. If you have ever seen the same smiling stock model on two rival sites, that is why.
Rights-managed licensing, and why it still exists
A rights-managed license is priced for a specific use. You state the medium, size, placement, territory, duration, and print run, and the price is calculated from those. Step outside them and you need a new license.
It sounds inconvenient, and for most people it is. It survives because it buys two things royalty-free cannot: exclusivity, where you can pay so nobody in your industry or region uses that image for a set period, and certainty about scarcity, which matters for a national campaign or a book cover where seeing the same picture elsewhere would be embarrassing.
If you are writing blog posts or building a small store, you will almost never need this. Knowing it exists is enough, so that when a licensing page starts asking for territory and duration you understand what it is doing.
Editorial use only, and what "editorial" actually means
An image marked editorial use only is licensed to illustrate news or matters of public interest. It is not cleared for advertising, promotion, merchandising, or anything else that markets a product, service, or brand.
Images end up in this category for concrete reasons:
- There is no signed model release from the identifiable people in it.
- There is no property release for a private building, a distinctive interior, or an artwork visible in the shot.
- It contains logos, trademarks, product packaging, or a recognizable trade dress.
- It was shot at an event whose organizers restricted commercial exploitation of images.
The common mistake is assuming that "my blog is editorial" solves it. What matters is the use, not the publisher. A news site's article about a football match can use an editorial photo of the stadium. The same site's sponsored post promoting a betting app usually cannot, even though it sits on the same domain. If the image is doing promotional work, editorial-only does not cover it.
Attribution, done properly and done once
When a license requires credit, the practical standard is to include four things where they are reasonably visible: title, author, source, and license, each linked where a link exists. Creative Commons refers to this informally as TASL. A workable credit line looks like:
"Harbour at dusk" by A. Okafor, on Flickr, licensed under CC BY 4.0
Rules that save arguments later:
- Name the license version. "CC BY" without a version is ambiguous, and the versions differ.
- Keep the credit near the image where it is practical: a caption is better than a credits page buried in the footer, though the license does allow crediting in a manner reasonable to the medium.
- Say if you modified it. CC 4.0 requires indicating changes if you made any. "Cropped from the original" is enough.
- Do not remove existing credit information from the file itself. Deliberately stripping rights-management data is treated as its own violation in some jurisdictions, separate from the copyright question.
That last point catches honest people, because many editors and export presets discard metadata by default. Before you publish a batch, run one file through the image metadata viewer and check whether the creator and copyright fields survived your pipeline. It reads the file in your browser and shows you exactly which IPTC and EXIF fields are still there.
Where license information lives inside the file
Images can carry their own rights information in IPTC metadata fields, and this is where a well-prepared file tells you who owns it. The fields most relevant to licensing are Creator, Copyright Notice, Credit Line, Web Statement of Rights (a URL pointing at the license terms), and Licensor details.
This is not just archival tidiness. Google reads IPTC rights fields, along with schema.org license and acquireLicensePage markup, to show licensing information alongside images in search. If you are the creator, filling those fields in is how you make your terms travel with the file, our guide to embedding copyright and contact details covers exactly which fields to write and what survives an upload.
If you are the user, absent metadata proves nothing. Plenty of legitimately licensed images have been stripped clean by a CMS resize step. Metadata is evidence when present, not exoneration when missing.
Found on Google Images is not a license
Search results are an index. Google shows you where an image appears, not who owns it or what you may do with it. A picture can appear on fifty sites, forty-nine of which are also using it without permission.
The Usage Rights filter helps and is worth using, but it works by reading what pages and files declare. If a site labels an image Creative Commons incorrectly, or copies a CC-labelled image it never had rights to in the first place, the filter passes that error straight through to you. Google's own guidance is to verify the license on the source page before using an image.
A few other things that are not licenses, despite being widely believed:
- Crediting the photographer without permission. Credit is a condition of some licenses, not a substitute for having one.
- Removing it if the owner complains. It reduces damages in some contexts. It is not a defense to the use that already happened.
- "It was on social media, so it is public." Posting to a platform grants the platform a broad license. It grants you nothing beyond the platform's own sharing features.
- Using only a small part, or under a size threshold. There is no universal percentage rule, and cropping to a recognizable portion of a photo does not shed the copyright.
Model and property releases are a different question entirely
Copyright answers "may I copy and publish this file?". Releases answer "may I use this person's or this place's likeness in this way?". You can hold a perfect copyright license and still have a problem.
A model release is a signed permission from an identifiable person allowing their likeness to be used, usually specifying commercial use. It matters most in advertising and promotional contexts, and the underlying rights vary by country: personality or publicity rights in some, data protection and image rights in others.
A property release is the equivalent for property: a distinctive private building, an identifiable interior, a recognizable pet, or an artwork inside the frame. Rules on photographing buildings visible from public space, sometimes called freedom of panorama, differ sharply between countries.
Two practical consequences. If you commission a photographer for your own product or store, agree in writing who owns the copyright and get releases for any staff or customers in shot, the same discipline that applies when you prepare listing photos of a property you do not own. And if you download a CC0 photo of a stranger's face and put it in an ad, the CC0 dedication has not solved your release problem, because the photographer never had that right to give away.
A checklist before you publish
Run this once per image. It takes about a minute and it is the whole post in practical form.
- Where did this come from? Name the source page, not the search results page.
- What exactly does the license say? Read the actual terms, including version number. Do not rely on a label someone else typed.
- Does my use fit? Commercial or not, modified or not, and does my adaptation trigger ShareAlike.
- Is anyone identifiable in it? If yes, and your use is promotional, you need a release, not just a license.
- What credit is required, and where will it appear? Write it before publishing, not after someone asks.
- Save the evidence. A dated screenshot of the license page, the receipt, or the email granting permission, filed with the image. Licenses on the web change and pages disappear.
One more category to watch, because it is unsettled rather than merely complicated: images generated by AI tools. Several jurisdictions, including the United States, take the position that output with no meaningful human authorship is not protected by copyright, which affects whether you can stop others from copying your generated image. Separately, each generator's own terms of service govern what you may do with what you produce. Both questions are moving, so check the current terms rather than something you read a year ago.
When you are unsure, the cheapest fix is almost always to pick a different image. There are millions of clearly licensed ones, and no photo is worth an argument you did not need to have.