Artificial Intelligence (AI) Generated Art - FAQ

Interested in using AI images in your artwork? Check out our helpful FAQs on the subject.

What is AI Generated Art?

An AI art generator is a tool that utilizes artificial intelligence (AI) and machine learning algorithms to create unique images based on a basic input. A person would type in a simple text phrase and the generator would create one or more images based on that input. 

How do AI art generators work?

The AI art generator is an algorithm that has been “trained” on a dataset of images. The dataset can include billions of images scraped from all over the web, many from publicly available websites like Flickr, Pinterest, Fine Art America, Getty Images, and Deviant Art. The algorithm analyzes the images and makes associations between the image’s visual content and its metadata — file name, artist name, keywords, tags, etc. The newly trained algorithm is then ready to take a text prompt from a user, compare that text prompt to its vast dataset and the associations it made during training, remix all of that information together, and output one or more new images. 

Can I protect my artwork from being used to train AI generators?

The only 100% fail-proof way to protect artwork from being used to train AI generators is to not post it online. However, that’s not always a feasible option.

Some companies offer an opt-out option. However, that option would only apply to that one generator and not any of the others that create datasets based on publicly available images.

Researchers at the University of Chicago have developed a tool that makes pixel-level changes to a work of art with the intent of tricking AI generators who analyze it. That tool can be found here: https://glaze.cs.uchicago.edu/index.html

Does SAQA endorse any AI art generators?

No. There are many AI art generators available, but SAQA does not recommend or endorse any specifically.  

However, Adobe has taken steps to address the ethical concerns surrounding AI generators in the creation of its own AI model. Firefly is trained only on images in the public domain and those uploaded to Adobe Stock, Adobe’s stock image library. Adobe gives its artists permission to opt out of having their work used to train Firefly, but artists who opt in are monetarily compensated. Firefly is being incorporated into Adobe’s software, which is accessible to Creative Cloud subscribers.

Does creating and using AI art break copyright law?

No. Creating and using AI art does not break copyright law, as long as you are the one who created the AI art in question.

Many countries, including the USA and those within the European Union, have denied copyright to AI-generated artwork on the basis that the artwork contains no element of human authorship — it was created by a computer, not a person, and therefore cannot be copyrighted. However, the United Kingdom does offer copyright to AI-generated art. Every country has its laws and stance on protecting AI art.

The copyright issue is currently being questioned through several lawsuits working their way through US courts and abroad. It’s each SAQA member’s responsibility to ensure the legality of using AI art within their own country of residence.

Is it okay to use AI art in artwork submitted to SAQA?

SAQA considers AI generators to be a tool in an artist’s toolbox for creating art. However, in keeping with SAQA’s Original Art policy, artwork submitted that uses AI must be transformed enough to make it uniquely the member’s creation.

Beyond SAQA’s policy, members using AI-generated art in their creations must ensure those generated images do not break copyright law. Additionally, it’s the responsibility of each member to read and follow the Terms of Service (ToS) and End User License Agreement (EULA) for each AI generator they are using. Some generators have restrictions on how the generated artwork they create can be used. Following an AI generator’s ToS and EULA is the responsibility of members choosing to use AI art in their work.