This first trade group in the sector of the AI dataset licensing industry, the Dataset Providers Alliance (DPA), consists of seven content-licensing sellers who specialize in providing music, image, video, and other datasets for training Artificial Intelligence (AI) systems.
With the leadership of industry pioneers, The DPA intends to support ethical data acquisition and safeguard the intellectual property rights of content owners.
"Ethical data sourcing" is at the forefront of the DPA's mission. The group is committed to ensuring that AI systems are trained using datasets that respect the rights of individuals included in the data.
In recent years, tech companies such as Google, Meta, and OpenAI have faced a string of copyright lawsuits due to their use of generative AI technologies which enable machines to mimic human creativity.
These companies have been training models by feeding them vast quantities of content, often without the consent of the creators or rights holders. The DPA intends to raise awareness and establish standards to ensure that licensed data is used in a legally compliant manner.
The Dataset Providers Alliance members are all prominent companies in the industry, including Rightsify, a U.S.-based music dataset provider, vAIsual, an image licensing service, Pixta, a Japanese stock photo provider, and Datarade, a data marketplace based in Germany.
Emerging Trends for New Industry And Calls for Legislation
The demand for licensed data is expected to grow significantly if copyright owners prevail in their legal fights. This has led to the emergence of a nascent industry, consisting of companies that package content and sell access to it for AI systems.
To establish ethical standards for this trade, various groups, including the nonprofit organization Fairly Trained, have been formed. Fairly Trained certifies models that have not used copyrighted materials without a license, ensuring fairness and legality in AI development.
The U.S. bill known as the NO FAKES Act, which was introduced last year, exemplifies the type of legislation supported by the DPA.
This bill aims to establish consequences for producing unauthorized digital copies of individuals' voices or appearances. By championing such legislative efforts, the DPA aims to establish a clear legal framework that protects the interests of all stakeholders involved in the AI dataset licensing industry.
Besides advocating for ethical data sourcing and the protection of intellectual property rights, the DPA group seeks to push for greater transparency in training data.
This involves backing needs like those described in the European Union's AI Act and the proposed U.S. bill, the Generative AI Copyright Disclosure Act. The DPA intends to release a white paper in July, detailing its stances on these crucial issues.
Alex Bestall, the CEO of Rightsify and its licensing subsidiary GCX, who led the establishment of the DPA, emphasizes that advocacy will play a crucial role in the alliance's activities.
Bestall highlights that resolving the ongoing battles in this field will take time and involvement from all stakeholders. By advocating for ethical standards and legislative measures, the DPA aims to foster a fair and balanced environment for AI dataset licensing.