AI images: To use or not to use?
Everywhere people are discovering for themselves the results of AI-generated images: uncomfortably good.
The potential benefits present a tantalising case: efficiencies gained from saving time on photoshoots and artwork, enabling illustrative images and creativity when none might otherwise be available, and closely aligning imagery with a brand’s identity.
But just because we can doesn’t mean we should.
While McDonalds thought their Christmases had come early when they released this AI-generated ad last December, they quickly discovered not everyone was as impressed with their AI skills as they were.
Burger King didn’t miss the opportunity to highlight just how much of a pandora’s box working with AI really is.
Know the risks
AI tools are advancing at such speed that they are far out-pacing industry norms and regulation, with policy largely yet to come.
It’s the Wild West in the meantime, but it doesn’t mean we should be cowboys.
One of the greatest risks is to the trust and integrity of your brand, especially with your customers and important stakeholders if they don’t like the way you have used (or failed to disclose) your use of AI images.
Take note that:
You can’t own an AI-generated image. At least, not right now. According to the US Copyright Office, images and videos entirely produced by AI aren’t eligible for copyright protection. It’s a legal can of worms, that’s only just cracking open but you should be wary that there’s nothing stopping others using your AI images if they want to, and the production of your AI images may have been trained using copyrighted content.
AI-produced images are always detectable. There are hidden watermarks in the code of AI-generated images, identifying them as AI even when it may not be apparent to the human eye. Increasingly, large platforms like Meta are detecting and auto-flagging content including AI.
Consumers are sceptical and cautious about AI-generated images. 84% want brands to disclose use of AI imagery, and 82% are open to their use in certain contexts. But 95% have concerns about AI image use overall (such as deception, lack of authenticity and ethics).
It won’t be the Wild West forever. While there may be an absence of rules and regulations today for AI image use, they will eventually catch up and this could have implications on any existing AI images you publish in the meantime.
Ask these questions when considering using AI images
We have put together a checklist of questions to help you make informed decisions about where and how your brand might use AI images:
What industry do we operate in?
Is our industry and are our stakeholders likely to be accepting of our use of AI tools in this way?
Are there stereotypes and biases in our industry we should be aware of? How do we ensure we do not negatively reinforce these stereotypes through AI imagery?
How important is place and people to our brand? Can these be authentically represented through AI generated images?
What type of images do we want?
Artwork that is clearly imaginative or illustrative may need less scrutiny than realistic depictions of people and/or products.
How do we intend to use the images?
Images used internally or in closed settings may need less scrutiny than those used in high profile, brand-representative spaces like your website and social media.
What could an alternative solution be?
Is an authentic photo/shoot achievable, and would it resonate better with our customers?
Failing that, is there suitable stock imagery to fit the brief?
How will we disclose AI-generated images?
Apply a solution that you are comfortable with for your brand (see note below on disclosure).
How will we monitor the images over time?
Ensure you have a process in place to regularly review suitability of the images and that they are complying with any policies that may emerge.
Make sure you file images clearly labelled as AI-generated, so anyone in the future is aware of their origin.
What feedback are we getting?
Consider how you can bring your customers and stakeholders on the journey, actively involving them in what works best for them.
How to disclose AI images
We recommend you find a place to proactively disclose your use of AI-generated images. With leading platforms developing their capability and policies to automatically flag and label these, it’s better if it comes from you.
When should you disclose?
If the image could be mistaken as a photo.
If the original image has been significantly altered and little of the original remains.
When you believe that it would be ethical to disclose.
How do you label AI-generated images?
At a minimum we recommend including a description of what the AI-generated image represents in the alt-text of the image (if online).
We also suggest you include a note somewhere that you have/are using AI-generated images, plus your reason why. This is where you could extend the invitation for feedback, showing you’re willing to listen and engage with stakeholders on the topic.
TLDR
Remember how fast AI tools and their use is evolving: It’s likely this guide will be out of date soon after it’s published. Continue to do your research and review your position on AI images and video.
The TLDR? Proceed with caution.