The Future of AI- Generated Images – Why It’s Both Fascinating And a Huge Legal Risk

July 21, 2022
Share on facebook
Share on twitter
Share on linkedin
Share on email

Get the latest posts delivered to your inbox


Why use JumpStory?

- Save time: We've curated the best images for you

- Save Money: Unlimited use for just $29 per month

Have you heard about Dalle 2 (Open AI), DALL·E mini and Google Imagen? They’re all exciting examples of text-to-image technology, in which images are created from text input.

At JumpStory, we love new technology but we also want to make sure that you’re aware of the legal aspects of these new opportunities. Especially if you’re an entrepreneur, advertising agency, freelancer or in another area of business, in which you need great images and videos for your work.

It’s truly fascinating. And it’s a lot of fun.

This is an example of Dalle-mini creating an image based on the text ‘Mark Zuckerberg as a ventriloquist’:

Mark Zuckerberg as a ventriloquist
Images like these have started going viral on Twitter and Instagram, and they are generated by the AI technologies that we mentioned above. Currently, Dalle 2 and Imagen seem to be the ones that are most advanced and create the most impressive results. Just have a look at some of these inspiring examples – all by Open AI:

A bowl of soup:

A bowl of soup that is a portal to another dimension as digital art

 

A bowl of soup that looks like a monster:

 

A bowl of soup that looks like a monster

 

Teddy bears mixing sparkling chemicals as mad scientists:

 

Teddy bears mixing sparkling chemicals as mad scientists

 

Shopping for groceries as a one-dine drawing:

 

Shopping for groceries as a one-dine drawing

 

DALL·E 2 can also make realistic edits to existing images from a natural language caption. It can add and remove elements while taking shadows, reflections, and textures into account. Here’s an example with a flamingo added to the photo:

 

flamingo added to the photo

 

It can even take an image and create different variations of it inspired by the original image. Here’s an example from a bathroom:

 

Different variation of bathroom

 

What the technology does is basically understand the relationships between an image and the text used to describe this image. The process is called ‘diffusion’, and it’s based on a principle, in which the AI starts with a pattern of random dots and then gradually changes this pattern into the final image – and recognizes specific aspects of the image in this process.

Learn more about how DALL-E 2.0 – also known as the Picasso of AI – works in this great article by Daniel Fein:
https://towardsdatascience.com/dall-e-2-0-explained-7b928f3adce7

You should also read this very interesting piece by Alberto Romero, in which Alberto explains the promise and limitations of this new and revolutionary form of AI:

https://towardsdatascience.com/dall-e-2-explained-the-promise-and-limitations-of-a-revolutionary-ai-3faf691be220

And finally, Michael Howe-Ely was one of the first people to get access to test the new and improved version. Check out his review here:

https://michaelhoweely.com/2022/06/21/testing-out-dall-e-2/

 

It’s fascinating to play around with but carries the same issues for commercial use

As fascinating as this new technology is to play around with, you have to be aware that it’s highly illegal to use it for any form of commercial purpose.

The companies behind the technology are doing their best to limit the ability of the technology to generate violent, hateful or sexually explicit images, but for you as a business user, this is not the main concern. It’s the lack of copyright and licensing for the actual use of these images for anything else than an article about the technology – such as this one.

If you go any further than writing about the technology, you’ve started using the images for purposes that are illegal, and you run a high risk getting sued. But why is that the case? As it states on the OpenAI website, the technology is experimental and meant for research only. The reasons for this are many:

  1. The dataset being used to train the technology has not been licensed for commercial use. What this basically means is that the images that are generated may be generated based on images that are not allowed to be used for commercial purposes. So even though your end result may not look anything like the original, you’re still violating the copyright and licensing terms.
  2. The output does not consider things such as model releases, landmark rights, etc. Basically that means you can risk generating an image that looks like, let’s say, The Little Mermaid, and the owners of The Little Mermaid trademark will then sue you – even though it’s not an actual photo of the mermaid.
  3. Read more about the many risks and limitations right here.

 

What should I do now?

As fascinating as this new technology is to play around with, you must be aware that’s it’s use is not a free-for-all. Our advice to you is simple: Play around with this amazing new technology but DON’T use it for your website, blog, newsletter, social media, pitch decks, etc. Create an account here on JumpStory when you need images, videos, or icons and want to make sure that you’re insured and are not running a legal risk.

We’ll finish up with an image generated by Dalle Mini. Summer Holiday is coming up, so we wrote the sentence:

Summer Holiday should be spent on the beach with the people that you love

Here is the result that we got:

 

dall.e-mini

Enjoy your summer!

 

About the author

Jonathan Løw is the co-founder of JumpStoryHe is one of Denmark’s most well-known entrepreneurs and business authors. He has been nominated as Entrepreneur of the Year and is amongst Denmark’s 100 most promising leaders according to a major Danish business newspaper. In addition to being a serial entrepreneur, Jonathan Løw is the former Head of Marketing at the KaosPilots – named Top 10 most innovative business schools in the world by FastCompany. He is also former Startup-Advisor and Investor at Accelerace – the leading investment fund for startups in Denmark.