ProRata tackles fractional monetisation of generative AI
The start-up claims to be able to identify the sources that have contributed to the generation of content by an AI. Several media and entertainment giants have already signed agreements with it.
Universal Music has become the first music major to sign a strategic partnership agreement with ProRata, a US start-up founded in 2024 that claims to have developed proprietary technology to analyse content output generated by an AI model and accurately identify the sources that contributed to its creation.
The company has developed a model for compensating contributors based on a concept of fractional attribution, which allows them to receive a share of the revenue generated by a derivative work in proportion to their contribution to the final result, based on a 50/50 sharing key with rights holders.
“We can take the result of generative AI, whether it's a text, an image, music or a film, and determine the origin of its components," says CEO and founder Bill Cross, an American serial entrepreneur known for inventing the pay-per-click model for search engine advertising in the late 1990s, whose patents were bought by Google1.
A technical challenge
ProRata, which is still awaiting validation of several patent applications, has so far not said much about the nature of the algorithmic methods used to achieve its aims. Precise identification of the source works is said to be technically very difficult, if not impossible in most cases, as generative AI models learn general statistical patterns from vast datasets, without retaining a direct memory of the training works2.
The startup has raised $25 million in Series A funding from incubator Idealab and investment funds Revolution Ventures, Prime Movers Lab and Mayfield Fund. The funds raised will enable the company to launch its own AI-based search engine later this year, demonstrating the effectiveness of its technology platform and fractional revenue model.
A number of media companies have already licensed their content to ProRata for this purpose, including the Financial Times and The Atlantic, the Axel Springer media group, Fortune magazine and more than 100 leading authors.
Bill Cross was also the founder of the Californian incubator Idealab, which managed to raise a billion dollars from major corporations in the early 2000s. This sum has enabled him to support the development of more than 150 new technology start-ups, around fifty of which have gone public.
A number of approaches are being explored to respond to this problem, such as the development of forensic analysis tools to detect similarities with existing works, the creation of reference databases of works used for training, or the exploration of 'watermarking' or traceability techniques built into AI models.