AI and Creativity. Where do we draw the line?
Creative
Creative

Recently, Thrive Creatives, Dave and Keiran joined a presentation for agencies to understand the legal side of using generative AI in client work, led by Albert Yung, the Associate Director of Advertising Law at the IPA. Here are their key take aways.
One of the biggest things that we took away was that it really solidified Thrive’s existing position on AI which is positive. We believe that AI should always be used as a tool in service of the final product, and not the final product itself, which feels like an important distinction. We still took loads of learnings away from it though.
Yes, AI can help us explore ideas, develop concepts, and speed up parts of the creative process. But if the data we’re inputting is being used to train the AI for anybody, and what we’re getting back could be drawing from work that exists somewhere else, can we really say that the idea it gives us is an original one? And that’s where things get complicated.
One of the questions asked, and a good one at that, was if we purchase a stock image and then use AI to create similar images in the same style, are we breaching any laws?
Or, if we purchase a stock image and use AI to bring that image to life through video generation, are we doing the same?
The answer isn’t necessarily straightforward. Even when we start with content we have legitimately purchased or created ourselves, there are still considerations around what the AI tool does with that input and what it creates as an output.
There’s also the uncomfortable reality that we may never really know whether an AI generated image, piece of copy or idea has replicated something that already exists. AI pulls from massive amounts of data, and we don’t have visibility of everything to know the answer.
Which brings us back to that original point, that we shouldn’t be expecting AI to give us a full or original idea. We should be using it as a tool to help us get to our own idea.
Copyright is an obvious consideration too, but it’s far from the only one. Albert talked through other areas including moral rights, passing off, trademarks and data privacy. One example really made this clear wherein a betting site had replicated the Angel of the North sculpture with AI and used it in an advert, to which Sir Antony Gormley, the sculpture’s creator, protested on the basis of his moral rights because he had not agreed to the work being used in this way. It’s a good reminder that just because something can be recreated, doesn’t mean it should be.
We should never be casually dropping confidential or NDA protected client information into an AI platform. That could create a breach in itself, and beyond the legal side of things, there’s also a simple consideration here: common sense. If a client has trusted us with information that isn’t public, we should treat it that way.
Another point that really made the room think was the question of whether AI content could manipulate an audience. Imagine an AI image that makes people believe a location, person or event is real when it isn’t.
Albert’s example was of an AI portrait of a homeless child being used to encourage donations to a charity. If that child doesn’t exist, are we manipulating the audience into believing something that isn’t true?
In the UK, there is currently no general law on requirement to label AI content. This led us think about the expectations we should set for ourselves as a design agency.
If you see an “AI content” label on a LinkedIn or Instagram post, is that something that makes you trust the content less? Or does it make you question why it needed to be disclosed and what part is AI? On the other hand, if something is AI but we don’t disclose it, is this then a question of ethics?
We don’t believe there is always an easy answer, however we do think it is something we should be actively thinking about.
For us, the biggest takeaway wasn’t that we should be using less AI, it’s that we should be using it with more thought. We should generally be assessing the benefits and risks of using AI, rather than simply asking whether it can do something and it’s something we should address collectively as an agency.
Thrive has a strong moral compass – openness, transparency and genuine creativity is something we value highly. Therefore, developing an AI policy that sets out how we use these tools, what we are comfortable putting into them, what needs checking, and where the boundaries are, makes sense for us.
We should all have a unified understanding rather than everyone making their own rules. Internally, a great way of policing that would be to appoint an “AI Champion”, someone people can go to when they have questions or aren’t sure whether something is appropriate.
Most importantly, whatever policy we create won’t just sit in a folder and gather dust. AI is shifting constantly and the tools and terms are changing, the law is evolving and our understanding of what is possible is moving with it.
For us, that’s probably the biggest lesson from the session: AI should help us make better work, but it shouldn’t replace the thinking, responsibility and judgement behind it.