OpenAI is fighting lawsuits from artists, writers, and publishers who claim their work was improperly used to train the algorithms behind ChatGPT and other AI systems. The company on Tuesday announced tools that appear to be aimed at placating creatives and rights holders by giving them some control over how their works are used by OpenAI.
The company says it plans to launch a tool in 2025 called Media Manager that will allow content creators to opt out of their work from the company's AI developments. In a blog post, OpenAI said the tool “allows creators and content owners to tell us what they own” and “how their work can be included in machine learning research and training. The company explained that it is a means to allow users to specify whether they want to be excluded or excluded.
OpenAI said it is working with “creators, content owners, and regulators” to develop the tools and intends to “set industry standards.” The company did not name its partners in the project or reveal exactly how the tool would work.
Open questions about this system include whether content owners will be able to make a single request that covers all their works, and whether OpenAI will allow requests related to models that have already been trained and launched. Includes whether or not. Research is underway into machine “unlearning,” the process of adjusting an AI system to retrospectively remove some contributions from training data, but the technology is not yet mature.
Ed Newton-Rex, CEO of Fairly Trained, a startup that certifies AI companies that use ethically sourced training data, said he welcomes the apparent shift to training data with OpenAI, but says implementation said it will be very important. “I'm glad that OpenAI is working on this problem. The details of whether it will actually help artists are not yet clear, but the details will be determined,” he says. The first big question that comes to mind for him is: Is this just an opt-out tool that will allow OpenAI to continue using their data without their permission unless the content owner requests an opt-out? Or is this an opt-out tool in the way OpenAI does business? Does this mean big changes? OpenAI did not immediately respond to a request for comment.
Newton-Rex is also interested in whether OpenAI will allow other companies to use the media manager, allowing artists to notify multiple AI developers of their preferences at once. “Otherwise, you're just adding more complexity to an already complex opt-out environment,” says Newton-Rex, the developer of the Stable Diffusion image generator and a former executive at Stability AI.
OpenAI is not the first to explore ways for artists and other content creators to signal their preferences for the use of their work and personal data in AI projects. Other tech companies, from Adobe to Tumblr, also offer opt-out tools for data collection and machine learning. His startup, Spawning, launched a registry called Do Not Train about two years ago, and creators have already added to his 1.5 billion favorites.
Spawning CEO Jordan Meyer said the company is not working with OpenAI on the Media Manager project, but is open to working together. “If OpenAI can facilitate registration and universal opt-out respect, we will happily incorporate their work into our suite,” he says.
