The recent frenzy over artificial intelligence sparked by the introduction of ChatGPT and DALL-E has reached a level where virtually everyone is aware of it—except for my late grandmother. The public reaction oscillates between overwhelming enthusiasm and anxiety about job security, showcasing a typical human response in the 21st century that often lacks both rationality and practical judgment. This situation is frustrating yet amusing. However, I anticipate that both sides of the debate will soon experience a slowdown in AI developments, and as someone immersed in technology, I believe this is a positive turn of events.

Recently, Meta and Shutterstock announced a collaboration where Shutterstock will provide its content to assist in training Meta's AI systems. This move is hardly surprising in a landscape where companies are eager to capitalize on AI, often producing what I refer to as "digital rubbish." It seems there's a constant need for generated content to complement the already abundant organic material, which raises questions about the ethics of such partnerships, especially with a company like Meta, known for either depending on subpar content or promoting its creation.

Shutterstock appears to want to position itself as a responsible player in this scenario, but their announcement from December 2022 about compensating contributors for AI rights raises concerns about the actual amount they will pay. Likely, the compensation will be minimal. Nonetheless, it's evident that at least one organization recognizes the importance of respecting intellectual property, as truly public domain image data is incredibly rare online—comparable to finding a live gazelle in a pride of lions.

Meanwhile, OpenAI has claimed that DALL-E was developed using GPT-3, trained on various text and image pairs. This assertion seems innocuous until one considers the origins of the data used. It's clear that the source material wasn't just invented out of thin air. OpenAI has not disclosed this data, and to my knowledge, they do not plan to do so anytime soon, which raises further questions about transparency. Based on the images produced, one could surmise that DALL-E utilized anything it could acquire from the internet, aside from images of celebrities and explicit content—at least we hope that's the case.

Stable Diffusion takes this concept even further. A friend of mine recently sent me a digitally altered image of Jenna Coleman, one of my favorite actresses, which looked remarkably real. He even printed it on a duvet cover! Just kidding… However, one might wonder, what's the harm in that? Unfortunately, there are significant issues here. Firstly, Jenna Coleman never consented to have an image generated of her. Additionally, the surge of AI-generated explicit images flooding adult sites raises serious ethical concerns. A task that once required hours of work from a skilled Photoshop user can now be completed in under a minute by someone with less noble intentions.

The glaring issue that underlies all these examples is the lack of permission and copyright considerations. Susie Kearley, one of my favorite writers on Medium, raises valid points about copyright violations in AI-generated images, but this is merely part of a larger narrative.

It's essential to recognize that artificial intelligence cannot exist without the foundation provided by human intelligence. Most commercial AI systems rely on data created by people in various forms. This includes everything from the text prompts used to generate images in DALL-E to the language employed to create articles with ChatGPT. This data is often repurposed for subsequent searches, but let's shift our focus to the vast array of content generated by individuals daily and published online.

Many may still find this shocking, but just because something is accessible on the web does not mean it falls into the public domain. For example, I hold all rights to the text in my articles. While you are welcome to read and draw inspiration from them, attempting to copy or use any part of it would result in legal action. I may be a nice person, but cross me, and I can become quite formidable.

A crucial distinction must be made between merely consuming content and utilizing it, as the latter carries significant legal ramifications.

A reckoning is imminent for AI companies once people realize their work is being exploited for free and subsequently sold back to them. Some platforms have already begun banning AI-generated content, which is a step in the right direction, but it only addresses part of the issue. The other critical aspect is restricting the use of content for AI training—my images, my articles, my software code. If Microsoft’s Co-Pilot seeks to utilize my work without permission, I will take action to protect my intellectual property, even if it means leaving GitHub.

The fruits of our labor cannot serve as a free resource for artificial creations. This is a form of exploitation. We cannot disregard copyright and licensing agreements in the name of AI advancement. Some argue that new laws are needed to address AI specifically, but I disagree. The existing licensing frameworks, despite their complexity, clearly indicate that "all rights reserved" applies universally, including to AI systems and their creators.

While AI presents exciting and genuinely useful possibilities, it must adhere to the same regulations as everything else. If it desires something, it must compensate the creators significantly. Increasingly, creators, writers, artists, and software developers will come to realize they are being taken advantage of by these impressive tools, which owe their capabilities to data they had no right to use.

We will soon witness class-action lawsuits against AI companies, and justifiably so. Utilizing data without consent is illegal; profiting from someone else's premium work—whether it's a free or paid product or service—is unlawful. Copyright lawyers will find themselves inundated with cases, working tirelessly to win settlements, likely earning more in a year than Wall Street executives do in a decade.

If AI companies believed this would be an easy venture, they are mistaken. Humanity always finds a way to push back.

Attila Vago — Software Engineer improving the world one line of code at a time. A lifelong tech enthusiast, coder, and advocate for web accessibility. Avid LEGO collector and craft beer lover! [Read my Hello story here!](https://example.com/hello) Subscribe and/or become a member for more insights on LEGO, technology, coding, and accessibility! For occasional readers, I also delve into various topics including writing and creativity.