AI systems and large language models require vast amounts of data to achieve accuracy, but they must avoid using data they don’t have rights to. OpenAI’s recent licensing agreements with The Atlantic and Vox illustrate the growing interest in securing AI-training content licensing deals.
Human Native AI, a startup based in London, is developing a marketplace to facilitate these transactions between companies developing LLM projects and those looking to license their data.
Aiding AI Training with Licensed Data
The goal of Human Native AI is to help AI firms find the necessary data for model training while ensuring rights holders are compensated. Rights holders can upload their content for free and connect with AI companies to negotiate revenue share or subscription deals. Additionally, Human Native AI assists rights holders in preparing and pricing their content and monitors for copyright violations. The startup earns a commission on each deal and charges AI companies for its transaction and monitoring services.
James Smith, CEO and co-founder, shared that his experience with Google’s DeepMind project inspired the idea for Human Native AI. DeepMind faced challenges with sourcing sufficient quality data for training. Observing similar struggles at other AI companies, Smith envisioned a solution.
“It feels like we are in the Napster era of generative AI,” Smith remarked. “Can we progress to a better era where content is easier to acquire and creators are compensated? Why isn’t there a marketplace for this?”
Smith pitched the concept to his friend Jack Galilee, an engineer at GRAIL, during a walk in the park with their children. Unlike previous ideas, Galilee encouraged pursuing this one.
Early Success and Funding
The company launched in April and is currently in beta. Smith noted that demand from both sides has been promising, with several partnerships already in the pipeline. This week, Human Native AI announced a £2.8 million seed round led by British micro VCs LocalGlobe and Mercuri. Smith plans to use the funding to expand the team.
“I’m the CEO of a two-month-old company and have secured meetings with CEOs of 160-year-old publishing companies,” Smith said. “This indicates strong demand from the publishing side. Similarly, discussions with major AI companies have been very positive.”
Although still in its infancy, Human Native AI seems to address a crucial need in the growing AI industry. AI companies require large datasets for training, and facilitating easy, controlled access for rights holders appears to be a mutually beneficial approach.
“Sony Music recently sent cease and desist letters to 700 AI companies,” Smith noted. “This highlights the vast market and potential customers for data acquisition. The number of publishers and rights holders could reach thousands, if not tens of thousands. That’s why infrastructure is essential.”
Smith believes that Human Native AI can be particularly advantageous for smaller AI systems that lack the resources to secure deals with major publishers like Vox or The Atlantic. The platform aims to level the playing field, enabling these smaller players to access training data.
“One of the main challenges with content licensing is the high upfront costs, which limit potential collaborators,” Smith said. “We aim to increase the number of buyers and reduce entry barriers, which is very exciting.”
Future Prospects and Ethical Sourcing
Looking ahead, Human Native AI plans to use the data collected to help rights holders better understand how to price their content based on historical deal data on the platform.
Smith also sees this as a timely launch. With the European Union AI Act evolving and potential AI regulations in the U.S., ethically sourcing data and being able to prove it will become increasingly important.
“We are optimistic about AI’s future and its potential benefits, but the industry must act responsibly to avoid harming the sectors that have brought us this far,” Smith said. “Ensuring inclusive participation is crucial. We are AI optimists, but with a focus on human interests.
See also: Apple Should Prioritize Practical AI Over Flashy Features
