Sony Music Accuses Tech Giants of Unauthorized Use of Songs in AI Development

Sony Music, the world’s largest music publisher, has issued letters to Google, Microsoft, and OpenAI, demanding transparency about whether they have used its songs to develop artificial intelligence (AI) systems. Representing artists like Beyoncé and Adele, Sony Music insists that no one is permitted to use its songs for training, developing, or profiting from AI without explicit permission.

In letters sent to over 700 companies, Sony Music expressed “reason to believe” that some recipients “may already have made unauthorized uses” of its music. The BBC has reached out to Google, Microsoft, and OpenAI for their responses.

Sony Music has imposed a deadline for these companies to respond and has vowed to enforce its copyrights “to the full extent permitted by applicable law,” including the EU’s upcoming AI Act.

AI Training Controversy

Sony Music’s grievance highlights a broader issue that has emerged with the latest generation of AI tools: the origin of the data used for training and whether permission was obtained. AI systems that can generate text, images, or music typically learn from vast datasets, raising questions about copyright and fair use.

Google and OpenAI have developed AI tools capable of generating music, though the specific training data they used remains undisclosed. Sony Music’s complaint centers on the notion that such AI tools might have been trained using millions of songs, including those owned by Sony Music.

Detailed Demands

In a detailed letter, Sony Music has requested that each of the hundreds of addressed firms provide:

  • A list of its songs used to train AI systems
  • Details on how these songs were accessed, such as via streaming services
  • Information on the number of copies made, their current existence, and the duration of their existence
  • Justification for the existence and duration of these copies, if any

Sony Music has indicated willingness to negotiate licensing agreements for future use but remains firm on the potential legal implications of past unauthorized uses.

Legal and Ethical Implications

The legality of training AI on copyrighted content without explicit permission is contentious. Both in the EU and the US, there is debate over whether such actions constitute copyright infringement or fall under fair use and “temporary copying” exceptions.

Nana Nwachukwu, a lawyer at the AI ethics firm Saidot, stated that under current EU rules, using copyrighted music to train AI “may constitute a copyright infringement.” However, exceptions exist for entities with “lawful access” to the music, such as those that are publicly accessible or licensed for AI training.

The forthcoming EU AI Act will enforce stricter regulations, mandating detailed documentation and transparency about the training data used by AI models. This includes public disclosure of training data summaries, adherence to copyright holders’ opt-outs, and compliance with EU copyright laws.

Ongoing Legal Battles

The issue of AI training data and copyright is being contested in courts across the US. Notable cases involve figures like “Game of Thrones” author George R.R. Martin, comedian Sarah Silverman, and the New York Times. In the music industry, Universal Music has taken legal action against AI company Anthropic, accusing it of infringing on the copyrights of hundreds of song lyrics.

As these legal battles unfold, Sony Music’s actions highlight the complex and evolving intersection of AI technology and intellectual property rights. zs zs zs zs zs zs zs zs