Suggestions
Ellery Kreloff
Co-Founder @ Artistree | Techstars '23 | Automating genAI compliance | Ex-HUMAN
Ellery Kreloff is an accomplished professional with a Bachelor of Science in Computer Science from the University of Wisconsin-Madison. With a diverse background in technology and data engineering, Ellery has held significant roles at reputable organizations such as Capital One and UW-Madison. As a seasoned Founder of Artifyc, Ellery has demonstrated a strong entrepreneurial spirit and leadership skills. During Ellery's tenure at Capital One, they served as a Senior Data Engineer, Associate Data Engineer, and Data Science Intern, showcasing a profound understanding of data analytics and engineering. Prior to this, Ellery gained valuable experience as a DoIT Web Development Intern and Service Desk Agent at UW-Madison's Division of Information Technology (DoIT), further enhancing their technical expertise. Additionally, Ellery's professional journey includes a Summer Internship at HUMAN, contributing to their well-rounded experience in the tech industry.
Highlights
BIG update on Andersen VS Stability AI! Judge Orrick allowed ALL of claims on copyright infringement to proceed! More on the order:
Midjourney's motion to dismiss Lanham Act (trade dress) claims was also denied. Discovery is going to be crazy!
Most interestingly, while DeviantArt was not alleged to have TRAINED any AI model, only implemented the tools provided by Stability and others utilizing LAION, its motion to dismiss the Copyright Act claims was denied.
This order implies that companies who utilize SD models for and/or LAION like datasets could now be potentially liable for copyright infringement violations, amongst other violations.
Full order here:
How soon will tech companies run out of training data for AI?
Sometime between 2026 and 2032, according to research group Epoch AI. Big tech companies are scrambling to protect their ability to monetize stolen copyrighted content, including the cringeworthy "Generate and Create" campaign.
Yet, if companies truly believe in the fair use defense, why have OpenAI, Meta, Google already set the precedent of paying for high-quality training data from Reddit, Newscorp, and Shutterstock (which pulled in a whopping 104M in 2023 from AI licensing)?
Just last month in Andersen v. Stability AI, Judge Orrick said that the copyright infringement claims made by artists likely had enough merit to continue, as: "plaintiffs have plausibly alleged facts to suggest compress[ed] copies... of their works are contained in the versions of Stable Diffusion identified."
Notably, his opinion did not address fair use.
