AI-generated content – the hidden legal dangers for SMEs

AI has become a topic that some people in business simply do not want to talk about, do not want to read about, and don’t really want to understand until it all feels like it has “settled down” a bit.

For that reason, the ongoing legal action from Disney and Universal against AI platform Midjourney seems like some sort of clash between giants in a far-off land, but in reality, the issues behind it affect us here at home.

AI is not going away. So, it’s important for businesses of all sizes to recognise that fact and at least take steps to understand the basics of how it works, and how to avoid it landing you in trouble.

For those small and medium businesses using AI tools to generate content, branding, or marketing materials, news of these massive legal cases should be a wake-up call on the dos and don’ts of experimenting with new tech.

Why this matters

Generative AI tools are fast becoming commonly used by small businesses. Whether it’s designing a logo, drafting meeting notes, or generating basic imagery, tools like Midjourney, DALL·E, and ChatGPT can be seen as quick, cost-effective solutions. But as with any aspect of business, if it looks too good to be true, it probably is.

In law, we can see one glaring problem with rapid AI content creation: if what you produce using AI mimics existing intellectual property, even unintentionally, you could be the one on the legal hook, not the AI company. Don’t be misled by headlines, even though an AI company is being taken to court for this specific case of Midjourney, that will not always be how things work going forward.

The Disney/Universal case centres on allegations that AI image-generation tools can produce output strikingly similar to well-known characters and franchises, and that the companies are aware of this. While this may sound like an issue for Silicon Valley, UK-based businesses should be paying close attention.

The same risks apply here, and the consequences can be just as serious.

Copyright and IP law: the Atlantic divide

One of the key issues often misunderstood by UK businesses is the difference between US and UK copyright laws.

In the US, “fair use” offers relatively broad and often flexible protection based on a balancing test regarding the purpose and character of the use, the nature of the work, the amount used and the effect on the market. In the UK, and specifically under Scots law, we operate under the narrower principle of “fair dealing,” which (although it is in some ways similar) is a more limited set of exceptions allowing use of copyrighted material without permission – such as parody, caricature and pastiche – provided the use is “fair” in terms of the amount of copyright material used, the necessity of using the copyright work for the purpose and the effect on the market for the original work used.  If the difference between the two tests seems unclear – well done, you’re well on your way to being a copyright lawyer.

Sometimes “fair use” and “fair dealing” will come to the same conclusion, but the US defence is considered to be more flexible and uncertain than the more predictable but likely narrower UK equivalent.

Importantly, the US test is more friendly towards “fair use” of copyright material in commercial contexts.

Commercial use in the UK can still be “fair dealing”, but the Courts will want to be sure that the use serves the permitted statutory purpose and that the use is proportionate, necessary, and doesn’t damage the market for the original.  The reasoning: if you’re making money from someone else’s work, fairness demands stricter justification.

This means that using AI to generate new content based on a famous character or brand is not protected just because it was created by a machine. If the AI-generated image resembles a Disney Princess or a Marvel character (for example) too closely, and you use it in a business context, you could be infringing on copyright even if you didn’t set out to do so.

Is it really “original” AI-generated content?

Most AI models, including Midjourney, are trained on vast data sets scraped from the internet. However, these companies rarely disclose exactly what content they’ve used.

To be clear, generative AI tools don’t delve into their memory banks to select something they’ve seen before.  But they might offer up a “remix” of it.

That means there’s no easy way for you, the business owner, to know if the AI is replicating something under copyright. It might look original to your eye, but legally it could be derived from protected material.

This legal grey area is dangerous territory for startups that may not have in-house legal teams or IP specialists on hand. You might think you’re safe because you’re not deliberately copying anything but in law, ignorance is not a defence.

Risks and reputational damage

Startups often use AI tools to build their brand identity quickly and affordably. But what happens if your new logo or marketing campaign is challenged by a global rights holder? The costs of rebranding, legal defence, and reputational damage can be severe and potentially business-ending for a small enterprise.

It’s also worth considering the indirect risk: platforms like Etsy, Amazon, or Instagram may take down your listings or profiles if they detect potential IP infringement. You may not get the chance to explain before the damage is done.

Practical steps for SMEs

So, what can you do to stay on the right side of the law while still benefiting from generative AI?

Avoid known characters and brands: Be careful when prompting AI and avoid asking it to create content “in the style of” or “like” specific copyrighted works or characters.

Get legal advice early: A quick consultation with a solicitor experienced in IP and tech law can save you much bigger headaches down the line.

Perform an image search:  This won’t guarantee that the AI tool hasn’t based the design on a pre-existing copyright work, but it helps you perform some due diligence.  If the artwork looks similar to something else on the internet, then be careful.

Use reputable tools: Stick to platforms that offer clear commercial licensing, allow you to opt out of using potentially copyrighted training data and review AI-generated output for unintended similarities to existing IP. If it looks familiar, it might be too close for comfort.

AI is a powerful tool for innovation and growth but it’s not a shortcut past copyright law. Understand the risks, stay informed, and don’t assume the big lawsuits overseas can’t affect you. Because in the eyes of the law, it’s often the user, not the toolmaker, who pays the price.

How we can help

If you have been affected by any of the issues discussed in this article or require legal advice, please contact us here or email Douglas McLachlan at Douglas.McLachlan@andersonstrathern.co.uk.

Article published in Digit News.