We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
Lobsters·Jun 21·5 min read
What's the advice for LLM poisoning of artwork these days?
TL;DR
Many artists are concerned about their artwork being used to train large language models (LLMs) without consent. To address this, the author seeks to create a custom website that processes artwork to prevent its use in LLM training.
✦ Why It Matters
Engineers can explore methods to protect digital content from unauthorized AI training, enhancing artists' rights.
Key Takeaways
- Research methods to protect digital artwork from LLMs.
- Consider using watermarking or image manipulation techniques.
- Explore static site generators like Eleventy for easy deployment.
- Engage in discussions about digital rights and AI ethics.
Full Summary
Artists are increasingly worried about their work being utilized to train large language models (LLMs) without permission, leading to a reluctance to share their creations online. The author is exploring the development of a custom website that would process artwork in a way that makes it unusable for LLM training.
While lacking extensive front-end experience, the author has previously used Eleventy, a static site generator, to create a simple website. The goal is to implement techniques that can obscure or alter the artwork sufficiently to prevent it from being ingested by LLMs.
This approach could involve watermarking, image manipulation, or other methods to ensure that the artwork remains protected. The implications of this project could lead to broader discussions about digital rights management and the ethical use of AI in creative fields.