Pictures are dangerous. Not the images themselves – the legal trap buried inside every photo found online. We learned that lesson the hard way. We received Getty Images demand letter by email. No DCMA takedown. No friendly notice. Just a demand. The fine was painful and permanent.
Every day, bloggers around the world grab photos from search engines and paste the photos into articles without a second thought. The assumption feels reasonable: photos on the internet are free to use. Wrong. Expensive, too.
US copyright law protects every photo the moment a photographer presses the shutter. No registration needed. No watermark required. Ownership is automatic. Agencies that hunt copyright violations – Getty leads every competitor by a wide margin – scan the web using AI tools that grow sharper every year. When the system finds a violation on your blog, no human reviews the case first. They just email you a bill.
This article covers what we know from real experience. Stock photos, screenshots, AI-generated images, press kits, free sources approved for commercial use – the rules differ for each category.
Two names dominate photo copyright enforcement online. Getty Images controls a vast share of the world’s licensed photography. The Associated Press enforces photo rights through a third-party agent called PicRights. Both agencies use scanning webbots to search for pictures that they own.
The Digital Millennium Copyright Act (DCMA) gives copyright holders the right to demand removal. Getty and PicRights choose not to use that right. Removal costs them nothing to demand – but a licensing fee pays their bills. The law allows a direct financial demand. So both agencies use the direct financial demand.
Is that fair? Your opinion is not a factor. It is legal and correct by US and EU copyright law.
Fighting is expensive. A copyright attorney costs more per hour than most settlement demands. Getty knows that math well. Settlement amounts sit just below the point where hiring a lawyer makes financial sense. Fighting may be worth considering when the image identification is wrong, when a valid license exists, or when the demand amount is very large. In most other situations, the math favors settling.
Ignoring a Getty or PicRights letter is a serious mistake. The demand does not expire. Agencies file suit against non-responders in US federal court. Statutory damages then climb well above the original settlement offer – reaching amounts that can seriously damage a small business. Courts have issued judgments in these cases many times.
Each image is a separate violation. Each carries its own fine. A letter citing several images may demand an amount that feels impossible. Negotiating is possible – agencies sometimes reduce multi-image settlements – but the liability for each image remains real and separate. Be aware however, that many people report in reducing the fine for multiple image enforcement. So if you are asked for a lot of money you may be able to settle for about 40% of the asking amount.
Paying one settlement does not stop the scanning. Automated tools continue crawling your pages every week. Every unlicensed image still on your site is a new exposure. Paying one fine without auditing the rest of your image library is like fixing one leak while the rest of the roof stays open.
Copyright is not the only legal trap in a photo. This article focuses on copyright only – but two other risks deserve a quick mention. Any recognizable face in a commercial context may require a model release, a signed legal document from the person in the photo. Children in any state of dress or undress carry liability that goes far beyond copyright law. Any image that places a real person in a false, embarrassing, or unwanted context may trigger a defamation or right-of-publicity claim. These are separate legal areas. Each one carries real consequences.
Three beliefs get bloggers into trouble every time. The first: the internet is free. – Not! Images are protected by copyright the moment a photographer takes them. The second: linking to the source makes photo use legal. Wrong – attribution is a courtesy, not a license. The third: downloading from a paid service grants distribution rights. Wrong – a personal license covers personal use only. Passing a licensed photo to another website, another client, or another publication almost always violates the original license terms.
A screenshot feels harmless. The legal reality is more complicated. Capturing a website as a whole image is generally acceptable under fair use – the screenshot illustrates the site, not the photo inside the site. The stock photo embedded in that screenshot is a different story entirely. Getty and similar agencies license photos to companies like Asana or Microsoft for use on their websites. That license does not transfer to anyone who takes a screenshot. The photo inside the screenshot is still searchable, still identifiable, and still owned. In some cases we use a tool to blank out the photographic part of a screen snapshot. But if the picture on the screenshot is integral to your product – like an AI face swapper – we will ignore the picture entirely and replace it with a stock photo from our library. That is why your screenshot may look different after we post the article.
Purchasing a license from any of these services covers your own use only. Sending a licensed image to another website – even to illustrate an article about your own product – almost always violates the original license terms. Each service link below leads directly to the license agreement.
These sources appear free or low-cost. Each one carries a specific risk that makes the images unsafe to publish on a commercial blog.
The company name was the first deception. Cyprus based Freepik – now rebranded as Magnific AI – built its entire brand around the word “free” while burying an attribution requirement in the terms that made virtually every use a violation. Free tier images require a visible credit link back to Freepik on every page where the image appears.
Almost nobody does that. Almost nobody knows that.
A paid subscription removes the attribution requirement but introduces a worse trap – every image license expires the moment the subscription lapses.
The receiving website has no license at all, paid or otherwise, because the license belongs to the account holder only.
Finally, Freepik’s AI-generated images may carry invisible watermarks subject to future claims by a Cyprus-based company operating entirely outside US and EU jurisdiction. Always remember to read the fine print.
Both Pexels and Pixabay operate under EU jurisdiction – regulated, accountable, and subject to serious commercial law. The licenses on both platforms are irrevocable and free for commercial use. They do not require attribution. If a claim letter arrives citing a Pexels or Pixabay image, it is just a spam attempt. We use the picture name to denote the picture source and that name is our proof of license.
One practical difference exists between the two platforms. Pexels image file names always contain the word “pexels” – the source is self-documenting. Pixabay “save as” file names do not always carry that same identifier, especially for older images. What we do is to add the word “pixabay” to the file name when saving. A file named “pixabay-business-meeting-JP4cu789a.jpg” is wasy to trace.
One caution on Pixabay: since 2023 the platform accepts AI-generated image uploads. AI generated images are not clearly without copyright.
The US Copyright Office ruled AI-generated images lack copyright protection. No human authorship, no copyright. The image enters the public domain the moment it is created. That ruling feels clean and simple. The reality is messier.
Consider the hours a skilled user may spend crafting prompts, refining outputs, and directing an AI tool toward a specific creative result. That process looks a great deal like authorship. Case law has not yet caught up to that argument – but case law moves slowly, and the argument is not going away.
Getty Images and Shutterstock – which Getty is currently in the process of acquiring – both operate their own AI image generators. Both claim their AI-generated images are commercially protected through subscription indemnification. Magnific makes the same claim. These companies believe something legally valuable exists in AI image creation. Their lawyers are already building the case.
Adobe Firefly, DALL-E, and other major AI tools now embed invisible provenance data directly into every image at the moment of creation. The standard is called C2PA. The watermark survives compression, screenshots, and re-encoding. A tool called SynthID embeds the signal at the pixel level – not in the metadata, inside the image itself. These marks identify the AI model that generated the image, the date, and the account that created it. Detection tools are already available. Better ones are coming.
An AI image that looks clean today may be fully traceable tomorrow.
For the purpose of our websites, we do not currently accept AI-generated images.
The question comes up: Feed a copyrighted image into an AI tool, adjust the tone, shift the curves, regenerate the pixels – and the original is gone. A new image exists. The logic feels sound. The reality is more dangerous.
SynthID – Google’s invisible watermarking standard now embedded in major AI tools – was specifically engineered to survive color shifts, compression, cropping, and moderate generative transformation. The watermark lives at the pixel level, not in the metadata. How deep a transformation actually destroys it is not publicly documented. That gap is deliberate.
Starting a generation with a copyrighted source image creates a derivative work. Getty does not need to prove the watermark survived. Getty only needs to prove the original image was the starting point. Derivative works require the same license as the original. The transformation argument does not erase that obligation – it adds a second layer of liability on top of the first.
AI-powered image search is improving rapidly. A transformation that defeats detection today may be fully traceable within a year. The risk does not stay fixed at today’s technical threshold – the threshold moves forward while your published image stays where it is.
Removing or defeating a watermark – even technically – shifts the legal category from accidental infringement to willful infringement. That distinction is not minor. Statutory damages for willful infringement can go as high as $150,000 per image for US based websites.
Nearly fifteen years of blogging teaches hard lessons. We have published thousands of articles across our network of websites. We have received two demand letters – one from Getty Images, one from AP via PicRights. We paid both fines.
At CompanionLink, our policy is simple: Every article on our network uses images sourced from Pexels, Pixabay (free), or Unsplash.
When a blogger submits picture without the URL, we ignore it and substitute a picture from our library. For efficiency we require the URL to the webpage – not to the image – as proof to validate the copyright. We do not post the URL. But we require it.
We also maintain a paid Shutterstock subscription for our own product imagery and select blog posts. Every image has a traceable origin. That traceability is not optional.
If you are sending an image to another website for publication – without a verified, transferable license – is not a neutral act. You will not see the consequences. The website that publishes your image does. A demand letter arrives at their address, not yours. The fine comes out of their revenue, not yours.
We ask you to take pictures seriously – this is important.
Questions Bloggers Ask About Pictures
No. Every photo is protected by copyright the moment a photographer takes the shot. No registration is required. No watermark is required. Finding a photo online does not make the photo free to use.
No. Attribution is a courtesy – not a license. Naming the photographer does not grant permission to publish the photo. A license is a separate legal agreement. Credit without a license is still infringement.
No. Linking to a source does not transfer any rights. The copyright owner did not grant permission by making the photo visible online. A link is not a license.
No. Google Images is a search index – not a photo library. Every image in the results belongs to its original creator. Google does not own the photos and cannot grant permission to use them.
Yes. Every photo carries copyright automatically. The only exceptions are images explicitly released into the public domain, or images licensed under Creative Commons with commercial use permitted. Assume every photo is protected unless the license clearly states otherwise.
Only for your own use. A personal license covers your own publications. Passing a licensed image to another website – even to illustrate an article about your product – almost always violates the original license terms. Each website needs its own license.
Sometimes. A screenshot used to illustrate a website as a whole is generally acceptable under fair use. A stock photo captured inside that screenshot is not. The photo inside the screenshot remains owned and searchable. We process all screenshots through an AI tool to remove embedded photos before publication.
With caution. The US Copyright Office ruled that pure AI-generated images are not copyrightable. However, platforms like Getty Images and Magnific claim licensing rights over AI images generated through their tools. AI images may also carry invisible watermarks traceable back to the source. We do not currently accept AI-generated images on our network.
No. A modified version of a copyrighted photo is called a derivative work. Derivative works require the same license as the original. Feeding a copyrighted image into an AI tool and adjusting the output does not erase the original copyright – it may add a second layer of legal exposure.
The demand does not expire. Agencies file suit against non-responders in US federal court. Statutory damages then climb well above the original settlement amount. Ignoring a Getty or PicRights letter is one of the most expensive decisions a small publisher can make.
No. Removing the image does not cancel the liability. Copyright infringement occurred the moment the image was published without a license. The demand covers the period of unauthorized use – not the current state of your website. Removal is the right first step. Removal alone is not a defense.
Probably yes – but the amount may be negotiable. Getty Images and PicRights skip the warning stage entirely and go straight to a financial demand. Removing the image does not cancel the obligation. Ignoring the letter leads to federal court and much higher damages. Fighting the letter requires a copyright attorney whose fees will likely exceed the settlement amount. Negotiating a lower figure is possible, especially for multiple images in one letter. The realistic options are pay, negotiate, or litigate – and litigation is almost never the right choice for a small publisher.
No – and attempting to do so creates a more serious legal problem. A modified version of a copyrighted photo is a derivative work. Derivative works require the same license as the original. Feeding a copyrighted image into an AI tool changes the pixels – it does not change the ownership. Invisible watermarking technology such as SynthID is specifically engineered to survive color shifts, compression, and AI regeneration. More importantly, defeating or removing a watermark shifts the legal category from accidental infringement to willful infringement. Statutory damages for willful infringement reach up to $150,000 per image. What started as a $1,000 problem becomes a business-ending one.
Being ready when they arrive with keys in hand, paperwork printed, and the car accessible…
The organizations that are building durable advantages are not running pilot programs — they are…
The engineering manager who turns strategy into execution, and the full-stack AI engineer who turns…
Professional fabricators recommend Hypertherm because they've run the machines, compared the consumable costs, experienced the…
Dubai in 2026 operates on a simple premise: time is the ultimate luxury, and every…
Reliable infrastructure, secure data pathways, and a commitment to user experience are the fundamental components…
