OpenAI has recently faced lawsuits from artists, writers, and publishers who claim that their work was used without permission to train the algorithms powering ChatGPT and other AI systems. To address these concerns, the company has announced the upcoming launch of a tool called Media Manager in 2025. This tool is designed to give content creators more control over how their work is utilized by OpenAI.
In a blog post, OpenAI described Media Manager as a way for creators and content owners to communicate what work they own and specify how they want it to be included or excluded from machine learning research and training. The company claims that it is collaborating with creators, content owners, and regulators to develop this tool, with the intention of setting an industry standard. However, specific details about how Media Manager will function have yet to be fully disclosed.
One of the key questions surrounding Media Manager is whether content owners will be able to submit a single request to cover all of their works, or if they will need to make individual requests for each piece. Additionally, it is unclear if OpenAI will allow requests related to models that have already been trained and deployed. The concept of machine “unlearning” is being investigated as a way to adjust AI systems to retroactively eliminate the influence of specific training data, although this technique is still under development.
Ed Newton-Rex, CEO of Fairly Trained, a startup that certifies AI companies using ethically-sourced training data, views OpenAI’s move towards addressing training data concerns as a positive step. However, he emphasizes that the success of Media Manager will ultimately rely on the details of its implementation. Newton-Rex questions whether the tool is merely an opt-out mechanism, allowing OpenAI to continue using data unless specifically excluded by content owners, or if it signifies a broader shift in the company’s approach to data usage.
OpenAI’s decision to create Media Manager reflects a growing trend within the tech industry to provide mechanisms for artists and content creators to express their preferences regarding the use of their work in AI projects. Companies like Adobe and Tumblr have also introduced opt-out tools for data collection and machine learning purposes. Spawning, a startup that launched the Do Not Train registry, has already seen creators register preferences for over 1.5 billion works. While Spawning is not currently involved in OpenAI’s Media Manager project, the company’s CEO, Jordan Meyer, is open to potential collaboration in the future.
OpenAI’s introduction of Media Manager represents a step towards addressing concerns over copyright infringement and data usage in AI development. The success of this tool will depend on its ability to provide transparency and control to content creators, while also setting a new standard for industry practices. As the AI landscape continues to evolve, collaborations between companies like OpenAI and startups such as Spawning may lead to more effective solutions for balancing innovation with ethical considerations in AI technology.
Leave a Reply