The EDA (Ethical Data Authority)

THE CHALLENGE

‍

Embracing Change in the Music Industry

‍

The rapid advancement of AI has created unprecedented opportunities for the music industry, but it has also introduced new challenges. The misuse of music data, copyright infringement, the potential for deepfakes and unauthorized voice cloning threaten the integrity of the industry and the livelihoods of artists.

‍

The US AI copyright act, EU AI Act, UK Artificial Intelligence (Regulation) Bill, and others focus on implementing transparency and permissions regarding training data for generative AI models. They all have similar criticisms and negative effects of slowing down innovation to deal with copyright clearance and regulation.

‍

OUR MISSION

‍

Protecting Artists, Empowering Innovation

‍

The Ethical Data Authority (EDA) is an industry-led initiative dedicated to addressing these challenges. By establishing clear standards and frameworks for music metadata, attribution, provenance and usage, the EDA empowers AI developers to create innovative products while protecting the rights of artists and copyright holders. Through open collaboration and industry leadership, we are building a future where music and AI coexist harmoniously.

‍

THE SOLUTION

‍

The EDA is solving some of these challenges by:

‍

• Engaging industry stakeholders, music rights organizations, data engineers and scientists, along with policymakers to collaborate on standards that will help facilitate ethical licensing, reporting and certification.
• Develop technical standards for certifying the source and history (or provenance) of AI generated outputs.
• Technical standards for metadata in AI rights, licensing terms and data ownership.
• Develop educational materials and resources to explain the importance of these standards and aid in adoption.
• Provide policymakers with research and reports on standards and practices that could help easily facilitate regulations.

‍

MAML (Music Attribution for Machine Learning)

‍

Ensuring Data Attribution, Integrity and Provenance

‍

MAML (Music Attribution for Machine Learning) is a technical standard for metadata attribution, certification, and history of music data sets. The initial core functionalities include:

‍

• Creating Content Credentials: Generating metadata that records the creation and modification history of a trained dataset.
• Verifying Content Credentials: Checking the authenticity and integrity of a data set.
• Displaying Content Credentials: Providing users with information about the content's provenance.
• Metadata Schema: Metadata for music rights attribution closely follows DDEX standard naming conventions.
• The SDK: Provides tools for developers to easily implement MAML standards and manage music data with enhanced transparency and security.

‍

HOW IT WORKS:

‍

1. Content Hashing:
   • When a dataset is created or modified, a cryptographic hash is generated from its data. This hash is a unique digital fingerprint of the content.
   • Any change to the content, no matter how small, will result in a completely different hash.
2. Data Credentials Creation:
   • Information about the data, such as creation date, author, and any modifications, is bundled into the Data Credential.
   • This credential is also hashed to create its own unique fingerprint.
3. Linking Content and Credential:
   • The hash of the content and the hash of the Data Credential are linked together, creating a cryptographic signature.
   • This signature is embedded within the Data Credential itself.
4. Verification Process:
   • When someone wants to verify the authenticity of the data set, they calculate the hash of the content again.
   • They then compare this newly calculated hash with the one stored in the Data Credential.
   • If the hashes match, it means the data set hasn't been altered since the Content Credential was created.

‍

The Role of Immutability:

‍

• Cryptographic hashes are inherently immutable. Once a hash is generated for a piece of data, it cannot be changed without altering the data itself.
• This property ensures that the integrity of the data and its associated metadata remains intact.

‍

GET INVOLVED

‍

Join the EDA and Shape The Future

‍

The EDA invites industry leaders, technologists, legal experts, and policymakers to join us in shaping the future of music and AI. By collaborating on the development and adoption of these standards, we can create a thriving ecosystem where innovation and ethical practices coexist. Together, we can build a foundation for trust, transparency, and responsible AI in the music industry.

‍

Beginnings of a Formal Structure:

‍

To ensure effective collaboration and standard implementation, the EDA is establishing the following core committees:

‍

• Standards Development Committee: Responsible for creating and maintaining technical standards for music metadata, attribution, provenance, and usage.
• Industry Engagement Committee: Dedicated to fostering partnerships and building consensus among industry stakeholders.
• Policy and Advocacy Committee: Focused on engaging policymakers and advocating for industry-friendly regulations.
• Governing Board: Responsible for establishing the EDA’s legal structure, governance policies, and membership criteria.

‍

The music industry needs solutions to combat regulations and red tape that would slow down AI innovation. We need to make it easier to train ethical AI models rather than have it slow down innovation. We need new frameworks to facilitate data set licensing, provide attribution, and certify authenticity. Join us in this critical endeavor by contributing your expertise, resources, and passion to the EDA.

‍

CONTACT US

‍

‍

‍