Intellectual property-focused Story Protocol is partnering with decentralized AI protocol Ritual to make AI models and outputs trackable and monetizable.
The pair hope to create sprawling intellectual property graphs where developers can see how their AI models are being deployed and fine-tuned. Additionally, they want to ensure that content produced by AI models is watermarked and tracked on-chain. It’s the latest attempt at integrating blockchain technology into the lucrative world of AI development.
AI models, like OpenAI’s ChatGPT, have rapidly improved in their ability to draw conclusions and generate content. However, their creators often lack information as to how and by whom the models are being used. Story co-founder Jason Zhao said the protocol is in talks with some AI developers who simply want to track all of the derivative models being spun off their core model. And the distribution of off-chain AI content can be generally difficult to control.
Read more: Funding Wrap: Decentralized AI is all the craze
Story Protocol makes creation and monetization involving intellectual property easier through blockchain tech, with this AI project being one of its early use cases.
For its part, Ritual helps inferences, or conclusions that AI models make when given new data, interact with smart contracts on-chain. It developed a software development kit (SDK) for its so-called Infernet that brings off-chain compute on-chain.
Through the Story-Ritual collaboration, developers will be able to set rules for how their AI models powered by Ritual’s Infernet are being utilized downstream, and also set terms for how the AI model’s outputs are used. For example, a developer could set a non-commercial license where no one can earn revenue from generative images created by a certain model, Zhao said.
“We’re enabling a network where models can be remixed and built on, and the rights, ideas and licensing of each developer’s model is tracked and protected on-chain,” Ritual co-founder Niraj Pant said in a direct message.
Blockchain’s potential for AI model development is an ongoing subject of interest in the crypto world. The models themselves wouldn’t necessarily go on-chain, but blockchain could help around the edges. Ethereum co-founder Vitalik Buterin enumerated some possible integrations in a blog post.
Among the more popular ideas, one approach involves leveraging blockchains to authenticate the content used for training AI models or generated by them. In January, FOX Corp teamed up with Polygon in unveiling a tool for tracing the origins of content being used to train AI models. Now Media released Sovereignty, a content verification tool on Base meant to combat misinformation in the “age of AI and fake news.”
Read more: What can blockchain do for AI? Not what you’ve heard.
Story is betting that as more content is created with the help of AI, a market will emerge for a protocol governing ownership of the models and outputs. Tokens are sometimes referred to as programmable money, where dollar-equivalents can be staked to secure blockchains or lent through DeFi protocols, Zhao said. According to Zhao, intellectual property is still stuck in its “dollar bill era.” Story hopes to make intellectual property similarly programmable, where content can be remixed or co-created without arduous legal processes.
“With all these powerful tools, almost god-like tools, and the god-like distribution of the internet, why is Hollywood on strike? Why are creators struggling to monetize?” Zhao said. “There’s a real need in AI for value capture, [intellectual property] provenance and attribution.”