AI Image Generation: Law & Ethics

Let me share my first personal experience with AI image generation before I understood the truth of what it is, how it works, and the many benefits that come when you finally embrace it: When MidJourney emerged in mid-2022, I was initially alarmed. I feared this new technology would drastically change my line of work. I anticipated that human artists would have to adapt to the #AITools or potentially leave the industry altogether.

Like many who were uneducated about the technology, I jumped on the fearmongering bandwagon. I openly supported groups calling for bans on AI image generation, believing it to be unethical and a threat to my career. In hindsight, I let my fear of being replaced cloud my judgment before taking the time to truly understand the facts.

This kind of knee-jeek reaction is something I despise in others, and I’m ashamed that I fell victim to it myself. I made my decision hastily, without properly educating myself on the capabilities and implications of this emerging technology.

Let’s Start With The Ethics:

Despite false claims by those who’ve decided what AI image generators are before doing proper research, they do not “steal art”. At least, no more than human brains do every moment without realizing it. That having been said, the human brain can process up to 10 images per second and store the equivalent of over 2.5 million gigabytes of data — far exceeding the 300 terabyte capacity of the world’s largest hard drive.

Everything people create, even in their dreams, is based to some degree on their life experiences and exposure to the creations of others — whether visual, auditory, or based on taste and touch.

The primary difference is that AI is less selective in how it manifests its imaginings, often requiring human prompting and editing to refine the output.

AI image generators work by using machine learning algorithms, specifically neural networks, to create images based on prompts and other input parameters.

Here’s a simplified breakdown of the process:

•Training: AI generators are trained by feeding them a massive amount of visuals and corresponding descriptions, which can include photos, art, video and audio, and miscellaneous digital assets, not much different from how human brains absorb countless elements through sensory experience.

•Neural Networks: The core technology behind these generators is neural networks, which are designed to mimic the way the human brain processes information. These networks learn to recognize patterns and features in the images they are trained on.

•Text-to-Image Translation: When you input a text prompt, the AI uses its learned knowledge to generate an image that matches the description. For example, if you type “a sunset over a mountain”, the AI will create an image that resembles this scene. This is how the human brain works while correlating memories.

•Stable Diffusion: The AI reverses the process of image dissipation to create coherent visuals. Simply put, the AI understands most of what it has begun generating as well as many of its flaws, repairing it so it better represents what it thinks we want to see.

•Fine-Tuning: Users can often adjust various parameters to refine the generated images, such as style, color, and composition. This allows for a high degree of customization and creativity. Anything the generator is not capable of doing can be edited in another program. Adobe Photoshop is my tool of choice.

And Now, The Legalities:

The claim that using AI platforms to generate images is illegal is inaccurate. In fact, there are no laws that prohibit or punish the use of AI for image generation. The only existing legal consideration pertains to copyright and ownership, which is clearly defined by the U.S. Copyright Office.

This law can be found in chapter 300, line 306 of the Compendium of United States Copyright Office Practices. It states, “The U.S. Copyright Office will register an original work of authorship, provided that the work was created by a human being”. In addition, according to the Federal Register, a division of the U.S. Copyright Office, “A work containing AI-generated material will also contain sufficient human authorship to support a copyright claim. For example, a human may select or arrange AI-generated material in a sufficiently creative way that the resulting work as a whole constitutes an original work of authorship.” and “A visual artist who uses Adobe Photoshop to edit an image remains the author of the modified image.”

While many AI image generators collect content from various sources, my personal favorite, Leonardo.Ai, operates differently. Instead, content and data is created by it’s members and shared among with others, who then customize and redistribute. This collaborative process generates an exponential network of creative exchange, which is quite wonderful.

So, while the use of AI image generation is a new and evolving practice, the legal landscape is clear — the technology itself is not illegal, as long as the resulting works adhere to existing copyright laws. Responsible use of these tools, including proper editing and attribution, is key to staying within the bounds of both the law and creative ethics.

Official Legal References: Copyright.Gov, Federal Register