AI systems and large language models require extensive data for accuracy but shouldn’t use data without proper rights. OpenAI’s licensing deals with The Atlantic and Vox illustrate both parties’ interest in securing AI-training content agreements.
Human Native AI, a London-based startup, is creating a marketplace to facilitate such deals between companies developing LLM projects and those offering data licenses.
Their goal is to help AI companies access training data while ensuring rights holders opt-in and are compensated. Rights holders can upload content for free, connecting with AI firms for revenue-sharing or subscription deals. Human Native AI assists in preparing and pricing the content and monitors for copyright infringements, taking a cut from each deal and charging AI companies for transaction and monitoring services.
James Smith, CEO and co-founder, drew inspiration for Human Native AI from his experience with Google’s DeepMind, which also struggled with data sufficiency. Observing similar issues in other AI companies fueled his idea.
“It feels like we are in the Napster-era of generative AI,” Smith commented. “Can we achieve a better era by facilitating content acquisition and providing creators with control and compensation? Why isn’t there a marketplace for this?”
He shared the concept with his friend Jack Galilee, an engineer at GRAIL, during a walk in the park. Unlike previous ideas, Galilee was immediately interested in pursuing it.
Launched in April and currently in beta, Human Native AI has seen promising demand from both sides, securing several partnerships to be announced soon. The startup announced a £2.8 million seed round led by LocalGlobe and Mercuri, two British micro VCs, with plans to use the funds to expand its team.
“I’m the CEO of a two-month-old company and have managed to secure meetings with CEOs of 160-year-old publishing companies,” Smith said. “This indicates strong demand from publishers. Conversations with major AI companies go similarly well.”
Although still in the early stages, Human Native AI seems to be addressing a crucial infrastructure gap in the AI industry. AI companies need vast amounts of data, and providing rights holders an easier way to collaborate with them, while retaining control over their content, is a strategic approach that benefits both parties.
“Sony Music recently sent letters to 700 AI companies demanding they cease and desist,” Smith mentioned. “This reflects the market size and potential customers seeking data. There could be thousands, if not tens of thousands, of publishers and rights holders. This is why infrastructure is necessary.”
Smith also noted the potential benefits for smaller AI systems lacking resources to secure deals with large publishers like Vox or The Atlantic, emphasizing the hope that Human Native AI will help level the playing field.
“One challenge with licensing content is the high upfront costs, limiting who can be engaged,” Smith said. “How can we increase the number of buyers and reduce entry barriers? This is very exciting for us.”
The future potential of the aggregated data on Human Native AI’s platform is another intriguing aspect. Smith noted that eventually, they will provide rights holders with better insights on pricing their content based on historical deal data.
The timing for Human Native AI’s launch is also strategic. With evolving regulations like the European Union AI Act and potential U.S. legislation, ethically sourcing data and maintaining proof will become increasingly important.
“We are optimistic about AI’s future but must ensure responsibility and avoid harming industries that got us here,” Smith said. “This is crucial for human society. We must find ways to enable participation. We are AI optimists advocating for humans.”
