Navigating the AI Act: Balancing innovation and intellectual property rights in the age of generative AI

History of the AI Act

The AI Act, previously known as the Artificial Intelligence Act, is a legislative act that was launched in August 2024 by the European Commission. The act was initially proposed in 2021 in response to the rise in popularity of generative AI systems, such as ChatGPT, however further amendments have been made to establish the EU’s regulatory and legal framework for AI, covering different types of AI across a range of sectors.

Whilst the UK does currently have its own AI regulatory framework, the EU’s AI Act still has significant impact on businesses within the United Kingdom that operate within the EU or it’s markets.

Purpose of the AI Act

The primary purpose of the AI Act concerning intellectual property is to create a regulatory framework that ensures the ethical and legal use of AI while balancing innovation and protection of Intellectual Property rights. The Act aims to clarify how existing IP laws apply to AI-generated content and inventions, and to address any gaps or ambiguities in the current legal framework.

Limitations of the AI Act

Whilst the Act underscores the importance of protecting IP rights, some of the specific details remain to be seen. It is likely the EU courts will be expected to provide further clarification on how copyright holders can effectively flag works for exclusion from generative AI use.

Obligations for generative AI providers

1. Transparency and Disclosure: Generative AI providers must disclose the sources of the data used for training AI models, particularly if the data includes copyrighted materials.
2. Compliance with IP Laws: Providers must ensure that AI-generated content does not infringe on existing IP rights and comply with Union copyright law. This includes implementing systems to detect and prevent the unauthorised use of copyrighted material in AI training and outputs.
3. Responsibility for Outputs: Providers are required to take responsibility for the outputs of their AI systems, ensuring that they do not violate IP laws or infringe on third-party rights.
4. Ethical AI Use: Providers must adhere to ethical guidelines in the development and deployment of AI, ensuring that the AI is used in a manner that respects IP rights and the broader legal framework.
5. Obtain Authority from Copyright Owners: Recital 105 of the Act outlines the importance for AI providers to obtain authorisation from rightsholders of copyright content used in training AI models, unless a copyright exception applies.

Considerations for IP owners and AI developers

IP owners should evaluate how AI-generated content fits into their existing IP portfolios, determining if new licensing agreements are necessary for such works. Simultaneously, AI developers must ensure they obtain proper permissions for any data used in training AI models, particularly when copyrighted content is involved. Both parties should be ready to adapt their legal strategies to the evolving IP landscape, which may include renegotiating contracts and updating terms of service to comply with the new regulations.

Practical Steps for IP owners and AI developers

1. Audit Data Use: Regularly audit the data used in AI development to ensure compliance with IP laws and the AI Act.
2. Update Contracts: Revise contracts and licensing agreements to address the use of AI-generated content and clarify ownership rights.
3. Implement Compliance Programs and Technological Solutions: Develop and implement compliance programs that ensure adherence to the new obligations under the AI Act. Both IP owners and AI developers should consider investing in technologies that can help track and manage the use of IP in AI training and outputs, such as watermarking or blockchain-based IP management systems.
4. Engage in Stakeholder Dialogue: Participate in ongoing dialogues with regulators, industry groups, and other stakeholders to stay informed and influence the development of future regulations.

Future Considerations for the AI Act

Considering the information we’ve reviewed, there are some changes on the horizon that could make it easier to comply with the AI Act. Examples of this include advanced watermarking which uses sophisticated algorithms to protect copyrighted information, reducing the chance of AI using copyrighted content e.g. covering faces of people within a photograph with an obstruction. Blockchain-based IP Management Systems can also be used to protect Intellectual Property by locking any data once entered into the blockchain, stopping it from being altered or deleted. This creates a temper-proof environment in which to track IP transactions and ownership.

Conclusion

It is imperative that providers of Generative AI prepare to ensure compliance with these new regulatory requirements, reassessing their strategies around data and content creation. By clarifying ownership issues, introducing new obligations for AI providers, and promoting ethical AI use, the Act seeks to create a legal environment that supports both innovation and the protection of IP rights in an AI driven economy.