The world of AI art is rapidly evolving, but with this progress comes new legal and ethical challenges. Tensions have flared between industry leaders Midjourney and Stable Diffusion, with Midjourney accusing its competitor of data theft.
According to AI enthusiast Nick St. Pierre, employees from Stability AI allegedly infiltrated Midjourney’s database last Saturday, stealing crucial training data. This data consists of paired image and text prompts, essential for teaching AI models to generate images based on user descriptions. The alleged breach is also said to have caused a 24-hour service outage for Midjourney.
Adding a layer of irony to the situation, both companies train their AI models on massive datasets of existing images, raising questions about copyright and data ownership in the realm of AI art. It’s essentially an accusation of digital piracy within an industry built on borrowed imagery.
David Holz, CEO of Midjourney, has confirmed ongoing investigations and expressed strong suspicions towards Stability AI. Emad Mostaque, CEO of Stability AI, however, denies any wrongdoing and has even offered to assist with the investigation. Interestingly, Mostaque is a known supporter of Midjourney, having previously invested in the project.
While the full details remain unclear, Holz has reportedly offered to share information with Mostaque to aid their investigation. Whether this signifies a potential truce remains to be seen.
Regardless, this conflict highlights the pressing need for clear solutions regarding legal and ethical issues within the rapidly developing field of AI art.