Can artists protect their work from AI? – BBC News
TLDRThe BBC News video explores the burgeoning conflict between AI art generators and artists. It highlights the case of Carla Ortiz, whose artwork was used without consent for AI training, leading to a lawsuit against AI companies. The segment introduces 'Glaze,' a new technology aimed at protecting artists' copyrights by subtly altering images so they mislead AI models but remain visually unchanged to humans. Despite Glaze's promise, skepticism remains about its long-term efficacy and the ethical use of AI in art.
Takeaways
- 🎨 AI art has significantly advanced, with one piece selling for over $400,000 at Christie's in 2018.
- 🖼️ AI image generators like Dolly and Stable Diffusion can create art in seconds by mimicking styles from ingested images.
- 🚫 Many artists have not given consent for their work to be used in AI image generators.
- 💡 Concept artist Carla Ortiz discovered her art was used in an AI dataset without her permission, which she found invasive.
- 📚 AI models learn from vast datasets of images and text, raising ethical concerns about art theft at an unprecedented scale.
- 📉 Carla Ortiz and other artists filed a class action lawsuit against AI image generators like Stability AI.
- 🛡️ Professor Ben Zhao and his team at the University of Chicago developed 'Glaze', a solution to protect art from being used in AI models.
- 👀 Glaze makes imperceptible changes to art that are significant to AI, preventing it from learning the artist's style.
- 🔍 Critics argue that AI art generators are inspired by studying pieces, similar to how humans learn from other works.
- 📋 Some artists are willing to work with AI, but they want an opt-in process, not an opt-out one.
- ⏳ Despite the development of tools like Glaze, there is no guarantee it will last forever against determined attacks.
- 🌐 The internet community is already attempting to bypass Glaze, highlighting the ongoing need for regulation and public awareness.
Q & A
What is the recent development in AI art that has garnered significant attention?
-AI art has made a significant leap with one piece selling for over four hundred thousand dollars at Christie's auction in 2018. This is due to image generators like Dolly and stable diffusion that allow almost anyone to create new art in seconds.
How do AI models learn to mimic styles, including those of specific artists?
-AI models learn to mimic styles through a process called training, where they ingest millions or even billions of images scraped from websites around the web, combined with text describing the images, thus creating a data set that enables them to generate almost any type of image from a simple text prompt.
Why is there a concern among artists about their work being used in AI image generators?
-Many artists have not given their consent for their art to be used in AI image generators. This has led to concerns about art theft on an unprecedented scale, as artists' styles are being replicated without their permission.
Who is Carla Ortiz and what did she discover about her artwork?
-Carla Ortiz is a concept artist from San Francisco who has designed art for Magic the Gathering and Marvel's Doctor Strange movies. She discovered that her artwork had been scraped into an AI image data set without her permission, which she found invasive.
What action did Carla Ortiz and other artists take in response to their artwork being used in AI image generators?
-Carla Ortiz and a group of other artists filed a class action lawsuit against Stability AI and other AI image generators to address the unauthorized use of their work.
What is the solution developed by Professor Ben Zhao and his lab to help artists protect their work from AI?
-Professor Ben Zhao and his lab at the University of Chicago have developed a solution called 'glaze'. Glaze exploits the differences in how humans and machine learning models perceive visual images, making changes to the artwork that are almost imperceptible to humans but significantly alter how a machine sees it.
How does the 'glaze' technique help artists protect their work from being used in AI image generators?
-By applying the 'glaze' technique to their artwork, artists can publish their work online with the assurance that if an AI model attempts to steal their style from the piece, it will learn an incorrect style and fail to mimic the original when generating new art.
What is the criticism against AI art generators from the perspective of those who argue they are taking inspiration?
-Critics argue that AI art generators are taking inspiration in the same way a human does, by studying other pieces and learning from them. They assert that the AI is not creating copies, but rather learning and drawing inspiration, which they believe should not be subject to legal action.
How are companies like Stability AI and Adobe responding to the concerns of artists?
-Stability AI has stated that their new generators will be opt-out, meaning artists will have to actively choose to have their work included. Adobe, on the other hand, says its new image generator, Firefly, has only been trained on images from its stock library, implying a more controlled approach to the use of artwork.
What is the current status of the 'glaze' technique in terms of its effectiveness and potential circumvention?
-While 'glaze' is seen as a promising tool for artists, it is acknowledged that people are already trying to break it and get around its protections. The hope is that it will buy artists some time until more robust solutions or regulations are in place.
What role do regulators, artists, and the public play in shaping the future of AI art?
-Regulators, artists, and an informed public are crucial in ensuring that AI art tools are developed responsibly and ethically. Their input can help create a balance between technological advancement and the rights of artists, ensuring that the creation of AI art is done with consent and respect for original work.
Outlines
🎨 AI Art Controversy and the Impact on Artists
The first paragraph discusses the recent advancements in AI art, particularly the high sale prices at auction houses like Christie's. It highlights the use of image generators like Dolly and Stable Diffusion, which can create art in seconds by mimicking styles of artists without their consent. The paragraph introduces Carla Ortiz, a concept artist from San Francisco, who discovered her work had been used in AI datasets without permission. This has led to a class action lawsuit against AI image generators. The paragraph also introduces 'glaze,' a solution developed by Professor Ben Zhao and his lab at the University of Chicago, which alters images in a way that is imperceptible to humans but changes how machines see them, potentially protecting artists' styles from being stolen by AI models.
🛡️ The Future of Artistic Protection and Regulation
The second paragraph focuses on the potential of 'glaze' as a tool to protect artists' work online. It emphasizes the hope that 'glaze' will provide artists with time to adapt and for regulations to catch up with the rapid advancements in AI art generation. The paragraph also touches on the debate over whether AI generators are stealing styles or merely taking inspiration, similar to how humans learn from studying other pieces. It mentions that some artists are willing to have their work used by AI, but they prefer an opt-in process. The paragraph concludes with the acknowledgment that while 'glaze' may not be a permanent solution, it offers a temporary measure of protection and a call for a more informed public and regulatory involvement to ensure that these technologies are developed ethically and with the consent of the artists involved.
Mindmap
Keywords
💡AI art
💡Training
💡Consent
💡Class action lawsuit
💡Glaze
💡Image generator
💡Opt-in/Opt-out
💡Regulation
💡Informed public
💡Intellectual property
Highlights
AI art has seen a significant leap, with one piece selling for over $400,000 at Christie's in 2018.
Image generators like Dolly and Stable Diffusion allow anyone to create new art in seconds by mimicking styles of specific artists.
AI models are trained on millions of images and text descriptions, enabling them to create images from text prompts.
Many artists have not consented to their work being used in AI image generators.
Carla Ortiz, a concept artist from San Francisco, discovered her art was used in an AI image dataset without her permission.
Carla Ortiz and other artists filed a class action lawsuit against AI image generators like Stability AI.
To avoid unauthorized use, Carla removed her art from the internet.
Professor Ben Zhao and his lab at the University of Chicago developed a solution called GLAZE to protect art from AI.
GLAZE makes imperceptible changes to art that significantly alter how machines perceive it.
Artists can use GLAZE to prevent their style from being stolen by AI models.
GLAZE demonstrates a way for artists to maintain control over their work online.
Critics argue that AI art generators are inspired by studying pieces, similar to how humans learn.
Companies being sued are asking for the case to be dismissed, claiming they are not making copies.
Some artists are open to using their work with AI, but they want an opt-in process rather than opt-out.
Stability AI and Adobe are moving towards opt-out models and using stock library images for training.
Adobe contributors have expressed concerns about the use of their images for AI training without explicit agreement.
People are already attempting to bypass GLAZE, but it is hoped it will buy artists time until regulations and public awareness catch up.
The pressure from regulators, input from artists, and an informed public are crucial for the ethical development of AI tools.