Beware the Lure of Easy Images – Seriously

By: Margaret Johnson

It’s time we had a heart-to-heart talk about this. We all know that images make web content look good, much like butter makes things taste good, right? Images make everything more compelling. We are fed studies about how social posts get more engagement when there’s an image in them, how presentations are more compelling with pictures, and how a text-heavy website won’t be as engaging as one broken up by attractive graphics.

Are we all agreed here? Images are good.

When Good Images Go Very, Very Bad

Well, yes – except when they are very, very bad.

You must be wondering… what could make an image bad?

I’m glad you asked. Any image can be bad if it results in you getting a nasty-gram from a lawyer, or a bill for $800 – or more – from an image service provider. Don’t you agree?

How a Movie Launch Revealed a Copyright Trap

When I first wrote this post, a movie release had just spawned a set of hilarious parody “stock photos” starring the film’s celebrities. You know the type: the obligatory group shot around a conference table, someone smiling in front of a chart with arrows headed up to success. Those images were all over Facebook for a week, and I trolled the comments on several of the posts. I was alarmed to see how many people were talking about using those photos on their websites or in ad campaigns.

Alarmed. I was alarmed. Did you catch that?

Because those images were licensed for editorial use only. Using one in an ad campaign meant violating the image’s copyright. And yes, that means a bill or a lawyer nasty-gram could absolutely follow.

Right-Click, Save, Get Sued: Image Copyright Rules Haven’t Changed

Here’s the part of this post that has not changed one bit: just because an image is accessible through a web image search does not mean you can legally use it.

We’ve all been there. You need an image to complete your blog post, you search, you find one you like, right-click, save-picture-as, done. Need an image for an email or a landing page? Grab one from the web.

Getty Is Still Watching

Don’t do it. It was wrong when I first wrote this, and it’s wrong now. Getty Images (most famously, but others too) built a reputation for finding improperly licensed uses of their images and sending legalese-filled letters to the violators. Even if you never see a courtroom, you could end up paying a hefty sum to an image owner, or to an attorney. And image-matching technology has only gotten better at finding unlicensed uses since then.

Rule of thumb: Unless you see language that says you CAN use an image, assume that you cannot.

Then AI Changed the Picture. Literally.

Now here’s the part I could not have imagined when I first wrote this post: today you can simply generate the image you need.

Need a photo-realistic image of a golden retriever wearing safety goggles in a wood shop, lit like a catalog shoot, in your brand colors? Type that sentence into an AI image tool and you’ll have options in seconds. No stock library subscription. No license tiers. No hunting through twelve pages of almost-right photos. You can produce precisely the image your content needs, instead of settling for the closest thing someone else already shot.

For the exact problem this post warned about, that’s real progress. Nobody at Getty is sending a nasty-gram over a picture of a dog that never existed.

The New Lure Looks Exactly Like the Old One

And that is the part that should make you sit up.

Free. Instant. Right there in the browser tab you already have open. No paperwork, no invoice, no waiting on a designer. Every single thing that made right-click-save so tempting is true of the generate button, except more so.

The old lure got you in trouble one way: you took something that belonged to someone else. The new lure has five ways, and none of them looks like trouble at the moment you click download. Here they are, so you can spot them before they spot you.

Trap 1: You Can Use It, But You Can’t Own It

A purely AI-generated image cannot be copyrighted in the United States. Copyright requires human authorship. The Supreme Court declined to revisit that rule on March 2, 2026, when it denied review in Thaler v. Perlmutter, so the human authorship requirement is settled for the foreseeable future.

And before you decide that your very clever prompt makes you the author, the Copyright Office addressed that in its January 2025 report. Prompts function as instructions that convey unprotectable ideas. Writing a better prompt does not make you an author any more than describing a photo to a photographer makes you the photographer.

In plain terms: you can use the image, you just can’t own it. If you generate the perfect hero image for your flagship landing page, nothing legally stops a competitor from using something identical. Human editing, arrangement, and combination with your own material can earn protection for those contributions. A raw generation belongs to no one. That’s worth keeping in mind when you’re thinking carefully about your flagship landing page—the elements that actually drive conversions need to be assets you control and can stand behind.

That matters more than it sounds. Your logo, your product photography, your original diagrams, those are assets you control. A generated image is a rental with no lease.

Trap 2: The Generator Can Hand You Someone Else’s Property

The model was trained on somebody’s work, and sometimes it shows.

Disney, Universal, and Warner Bros. are all suing Midjourney over outputs that reproduce their characters, Darth Vader and Superman and Bart Simpson among them. In the United Kingdom, Getty’s copyright case against Stability AI mostly failed in November 2025, but the court did find trademark infringement where generated outputs carried a recognizable Getty watermark. In April 2026 a US court let Getty’s trademark claims move forward here as well.

Read that last part again. The watermark showed up in the output. Nobody asked it to.

Those cases are about the tool makers. But when a generated image lands on your landing page, in your email, or in your paid campaign, you are the one publishing it. If it carries a character, a logo, a watermark, or a face that belongs to someone else, “the AI made it” is not a defense anyone wants to test on your marketing budget.

What to Do Before You Publish Any Generated Image

Rule of thumb: Generate original scenes and subjects. Never prompt for real people, real brands, or a living artist’s style, and look hard at what came back before you publish it.

Trap 3: Your Tool’s Terms Are Not What You Assume

This is the one that catches growing companies.

Take Midjourney as an example. Company grosses more than $1 million a year? Your team needs the Pro or Mega tier for commercial rights, no matter which plan the person doing the generating happened to sign up for. Everything you generate lands in the public gallery, prompts included, unless you are on a plan with Stealth mode. And Midjourney offers no indemnification. You carry that risk yourself.

Other platforms sit in very different places. Adobe trained Firefly on licensed and public domain material and offers IP indemnification to enterprise customers. OpenAI and Microsoft offer indemnity on business and enterprise plans. Free and consumer tiers are usually excluded from all of it.

Two minutes reading your tool’s commercial use terms beats an unpleasant surprise, especially before you point a whole campaign at it. And if you have people on your team generating images for company use on their personal accounts, you have a version of this problem right now.

Trap 4: Now the Law Wants a Label

Here’s the newest wrinkle, and it arrived while most marketing teams were not watching.

New York’s synthetic performer law took effect on June 9, 2026. If an advertisement uses an AI-generated performer, a digital human who appears to be a person but is not any identifiable real one, the ad needs a conspicuous disclosure. Penalties run $1,000 for a first violation and $5,000 for each one after.

In the European Union, the AI Act’s transparency rules kicked in on August 2, 2026. Providers have to mark generated content in machine-readable form, and anyone deploying a deepfake, meaning content that looks like real people, places, or events, has to disclose it. Obviously fantastical images are not the target. Realistic-looking business photography is exactly the target.

And the FTC’s 2024 rule on fake reviews and testimonials already prohibits testimonials from people who do not exist, which is precisely what an AI-generated customer is.

None of this bans AI images. It bans passing off synthetic people as real ones without saying so. If your campaign features generated humans who look like customers, employees, or spokespeople, you have a disclosure question to answer, and it is now a legal one rather than a matter of taste.

Trap 5: Your Buyers Can Tell, and They Hold It Against You

Every trap above is legal. This one is worse, because it costs you the thing you were trying to buy with the image.

Start with your actual audience. TrustRadius surveyed roughly 2,500 technology buyers and vendors in January 2026 for its B2B Buying Disconnect report, and the finding that should stop you cold is this: 47 percent of buyers say they trust online resources less than they did a year ago, up from 39 percent. Vendor marketing materials came in last on credibility. Demos, free trials, prior experience, and user reviews came in first. Buyers are also checking everything, with 94 percent fact-checking what AI tells them and 72 percent doing it always or very often.

Gartner found the same instinct from a different angle. In a survey of 645 B2B buyers, 69 percent said they turn to a sales rep to validate what AI told them, and 51 percent said they are more likely to run into misleading information from generative AI than from any other source.

Put those together and the picture is clear. Your buyer arrives already suspicious of anything that looks machine-made, and already in the habit of verifying. That is the person you are about to show a generated photo of a team that does not exist.

Research on images specifically is still mostly consumer research, and it is not flattering. NIQ measured how people respond to AI-generated advertising using EEG readings alongside surveys. Viewers spotted the AI ads and disliked them. The words that came back were annoying, boring, and confusing, and the feeling did not stop at the ad. It attached to the brand. Getty’s survey work across tens of thousands of adults found nearly 90 percent want to know whether an image was made with AI.

The public examples write themselves at this point. Guess ran AI-generated models in Vogue and spent the following week discussing it. Coca-Cola’s AI holiday spot had trucks that changed shape mid-scene. In both cases the story became the images instead of the product.

Now put it back in your world. Your buyer is evaluating whether your company is real, competent, and safe to hand a budget to. An obviously generated smiling team, a conference room full of people with slightly wrong hands, a “customer” who does not exist, every one of those quietly answers a question you did not want asked, in front of an audience that already told researchers it trusts vendor material least of all.

AI Images vs. Stock vs. Real Photography: Your Options Compared

Your option When this was written Now
Grab from image search Illegal and risky Still illegal and risky
Free images with attribution Legal but ugly credits required Same, plus better free libraries
Buy stock images Safe, $1 to hundreds each Still safe, still generic-feeling, and increasingly assumed to be AI anyway
Hire a designer or photographer Best results, biggest budget Still the gold standard for anything that represents you
Generate with AI Didn’t exist Precise, fast, nearly free, can’t be owned, sometimes must be labeled, and your buyers notice

Proof vs. Concept: The Question That Determines Which Image to Use

We think the whole decision comes down to one question, asked before you go looking for any image at all:

Is this image doing a job of proof, or a job of concept?

A proof image carries evidence. It shows a real thing: your product doing the thing you say it does, the people who answer the phone, your actual office, a customer at their actual desk, a real screenshot with real numbers in it. Its entire value is that it is true. Generate it and you have destroyed the only thing it was there to do.

A concept image carries an idea. The abstract header at the top of a blog post. The visual metaphor in a slide. The decorative break between sections of a long page. Nobody believes that dog is a real employee. Its value is attention and pacing, not evidence.

Here’s how that sorts out in practice.

The job of the image Best source Never
Proof: product in use, screenshots, results Real capture from your product Generated, ever
Proof: your team, your office, your customers Real photography, even imperfect Generated people
Trust: testimonials, case studies, reviews The actual human, with permission A synthetic face or invented customer
Concept: blog headers, section breaks, metaphors AI generation or licensed stock Anything grabbed from search
Brand: logo, core visual identity, signature assets A designer, work you own outright AI output you cannot protect

The sorting principle is the same one that has run through everything we write about marketing automation: the closer a piece of content sits to the buying decision, the more it has to be true. Nobody converts because of your blog header. People do convert, or walk away, based on whether your product and your company look real.

AI Images in B2B Emails, Landing Pages, and Sales Enablement

Three practical places this bites in a B2B program.

Landing pages. The hero image on a gated landing page is usually working as proof, whether you intended it that way or not. A prospect who lands there is deciding whether to give up a phone number. A real product screenshot answers that question. A generated business scene answers a different question, and the answer is not the one you want.

Email. Your nurture emails have a harder job than any web page, because you showed up uninvited. Generic imagery already gets ignored. Generated generic imagery gets ignored and, when a reader notices what it is, drops a tiny mark against your name in the column where you were trying to build a relationship. That mismatch between what your email promises and what your content delivers compounds the problem—if you want to understand how those disconnects quietly kill response rates, the breakdown of how mismatches could be torpedoing your marketing efforts is worth a look. Most of the best nurture emails we see barely use images anyway. They use a person’s name, a sentence that proves you understand their problem, and one link.

Sales enablement. Whatever your team forwards to a prospect gets read at close range, sometimes in a room where somebody is looking for a reason to say no. That is the last place for a picture of nine people who do not exist.

Meanwhile, your marketing automation tells you which of these is working. If two versions of a landing page are in play, one with a real screenshot and one with a generated hero, you do not have to take our word for any of this. Track it. Watch the form completions. That is the argument to bring to whoever wants the whole site generated by Friday.

The One Thing AI Images Can Never Give You: Authentic Proof

The most valuable images on your website are still the ones AI can’t make. Real photos of you, your team, your office, your products, and your customers carry a trust signal no generated image can fake. In a web filling up with synthetic pictures, an authentic photo of actual humans doing actual work stands out more every year, and it’s the visual version of expertise no competitor can copy.

So use AI images the way you’d use stock: for the illustrative filler. Use real photos for anything that represents YOU.

How to Use AI Images Safely: A Practical Summary

When I first wrote this, the safe route meant creating your own images or paying for licensed ones, and the easy route could get you sued. Today the easy route is easier than ever, and it is no longer one trap. It is five smaller ones, spread across ownership, infringement, tool terms, disclosure law, and your buyer’s gut reaction.

None of that means don’t use AI images. It means know which job the image is doing, generate only for the concept work, read your tool’s terms once, never prompt a real person or brand, disclose when the law says to, and spend your real money on the photography that proves you exist.

And still, always, forever: stop right-clicking images out of search results. Some things never change.

Frequently Asked Questions

Can I legally use an image I found through a web search for my website or marketing?

Just because an image is accessible through a web image search does not mean you can legally use it. Getty Images (most famously, but others too) built a reputation for finding improperly licensed uses of their images and sending legalese-filled letters to the violators. Rule of thumb: Unless you see language that says you CAN use an image, assume that you cannot.

Can I copyright an AI-generated image I created with a detailed prompt?

A purely AI-generated image cannot be copyrighted in the United States. Copyright requires human authorship. Prompts function as instructions that convey unprotectable ideas. Writing a better prompt does not make you an author any more than describing a photo to a photographer makes you the photographer. In plain terms: you can use the image, you just can't own it.

Do I need to disclose when I use AI-generated images in my ads?

New York's synthetic performer law took effect on June 9, 2026. If an advertisement uses an AI-generated performer, a digital human who appears to be a person but is not any identifiable real one, the ad needs a conspicuous disclosure. Penalties run $1,000 for a first violation and $5,000 for each one after. In the European Union, the AI Act's transparency rules kicked in on August 2, 2026, and anyone deploying a deepfake has to disclose it.

What are the risks of using AI-generated images in B2B marketing?

47 percent of buyers say they trust online resources less than they did a year ago, up from 39 percent. Vendor marketing materials came in last on credibility. NIQ measured how people respond to AI-generated advertising using EEG readings alongside surveys—viewers spotted the AI ads and disliked them. The words that came back were annoying, boring, and confusing, and the feeling did not stop at the ad. It attached to the brand.

How can I use AI-generated images safely in my marketing?

Know which job the image is doing, generate only for the concept work, read your tool's terms once, never prompt a real person or brand, disclose when the law says to, and spend your real money on the photography that proves you exist. Use AI images the way you'd use stock: for the illustrative filler. Use real photos for anything that represents YOU.

Written by: — Chief Marketing Strategist

Margaret Johnson is a strategic thinker with a knack for getting to the root of challenges and helping to solve them. Devoted to providing education, knowledge, and ideas that help organizations thrive, she works with small and midsized businesses to drive revenue through effective sales and marketing, lead generation and nurturing programs, content creation, and strategic planning – and, in one example, has used her proven techniques to help an IT services organization grow from four million in revenue to nearly 16 million in revenue. An award-winning marketer, Margaret is also an effective and accomplished writer, speaker, presenter, coach, mentor, and collaborator.