The European Union's AI Act will establish a unified regulatory framework for AI across the EU, introducing significant new requirements for generative AI. Notably, it mandates the disclosure of copyrighted works used in training these systems. Here's an overview of what this means for AI developers and providers.
Background on the AI Act Amendments
The amendments to the AI Act specifically target generative AI, mandating providers to publicly disclose details of copyrighted works used in training their models. This move addresses growing concerns about copyright infringement in the training processes of generative AI systems. These systems, capable of creating text, images, and other media, have seen explosive growth and investment, exemplified by OpenAI’s ChatGPT and Microsoft's significant investments in AI.
Often, these AI systems utilise vast amounts of copyrighted material, sometimes scraped from the internet, leading to allegations of infringement from rightsholders. The new disclosure requirements could increase the legal scrutiny AI developers face, particularly in the EU, as they must transparently document their use of copyrighted materials.
New Obligations for Generative AI Providers
- Policy for Compliance: Providers of general-purpose models must establish policies to comply with Union copyright law, including honouring opt-outs from the EU's commercial text and data mining (TDM) exception using state-of-the-art technologies.
- Global Scope: The AI Act requires compliance with EU copyright law even if the model is trained outside the EU. This ensures a level playing field by preventing providers from gaining a competitive advantage through lower copyright standards outside the EU.
- Disclosure Requirements: Providers must publicly disclose details of the content used for training their models. This includes listing main data collections or sets, such as large private or public databases, while protecting trade secrets and confidential business information. No exceptions are made for open-source models, and those fine-tuning general-purpose models must disclose new training data sources.
Broader IP Considerations
The AI Act aims to balance innovation with the protection of fundamental rights, including IP rights. The specifics of achieving this balance remain to be seen, but the legislation emphasises protecting the intellectual property of AI system developers, such as trade secrets, while mandating public disclosure under certain conditions. Suppliers must provide necessary information for compliance without compromising their IP or trade secrets, and authorities applying the AI Act must also protect these interests.
Impact on the UK and Future Developments
Post-Brexit, the UK's approach to AI regulation has diverged from the EU’s. The UK has not implemented the EU's text and data mining exceptions, limiting such exceptions to non-commercial research. While the UK Government is consulting on broader exceptions, it has yet to propose obligations for AI providers to disclose copyrighted works used in training.
The UK aims for a lighter regulatory touch to foster innovation within the AI space. However, this approach may face challenges in a globalised market where the EU's stringent regulations could set a de facto standard. The UK's narrower exceptions and different regulatory frameworks could impact its competitiveness in AI development compared to the EU.
Practical Steps for AI Developers in the UK
UK-based AI developers should collaborate closely with IP lawyers to navigate the complex landscape of training data and copyright compliance.
Key steps include:
- Identifying necessary works and applicable copyright restrictions.
- Evaluating the extraction of protected data and potential infringement risks.
- Considering jurisdiction-specific strategies for training and hosting models.
- Implementing safeguards like filters and human oversight to prevent infringing outputs.
- Documenting compliance efforts meticulously to adhere to regulatory requirements.
Conclusion
- The EU AI Act represents a significant step towards regulating the use of generative AI and protecting intellectual property (IP) rights.
- Providers must adapt to these new requirements, carefully considering where and how to train their models to ensure compliance with EU laws. This will involve detailed documentation, transparent disclosure of training data, and robust compliance policies to navigate the complex landscape of AI and IP.
- The UK's different regulatory approach adds another layer of complexity, emphasising the need for thorough legal guidance and strategic planning in AI development and deployment.
For assistance in navigating these intellectual property matters, contact ip21 today.
