The AI Act and IP

The AI Act was agreed today. Here are the key points for intellectual property.

Generative AI and copyright

A general-purpose model is defined under Article 1(63) as:

an AI model, including where such an AI model is trained with a large amount of data using self-supervision at scale, that displays significant generality and is capable of competently performing a wide range of distinct tasks regardless of the way the model is placed on the market and that can be integrated into a variety of downstream systems or applications, except AI models that are used for research, development or prototyping activities before they are released on the market

This seems to include what is commonly known as “Generative AI”. Recital (105) says “General-purpose models, in particular large generative models, capable of generating text, images, and other content, present unique innovation opportunities but also challenges to artists, authors, and other creators and the way their creative content is created, distributed, used and consumed”.

Providers of general-purpose models must have “a policy to comply with Union copyright law”, including complying with opt-outs from the EU’s commercial TDM exception (Article 4(3) of Directive (EU) 2019/790) using “state of the art technologies” – Recital (105) and Article 53(1)(c).

Radically, it appears Union copyright law must be followed even where a general-purpose model is trained outside of the EU. Recital (106) says:

Providers that place general-purpose AI models on the Union market should ensure compliance with the relevant obligations in this Regulation. To that end, providers of general-purpose AI models should put in place a policy to comply with Union law on copyright and related rights, in particular to identify and comply with the reservations of rights expressed by rightsholders pursuant to Article 4(3) of Directive (EU) 2019/790. Any provider placing a general-purpose AI model on the Union market should comply with this obligation, regardless of the jurisdiction in which the copyright-relevant acts underpinning the training of those general-purpose AI models take place. This is necessary to ensure a level playing field among providers of general-purpose AI models where no provider should be able to gain a competitive advantage in the Union market by applying lower copyright standards than those provided in the Union.

Providers of general-purpose models must disclose details of the content used for training – Recital (107) and Article 53(1)(d). No exception is made for open source models – Recital (104). Providers fine-tuning general-purpose models need to disclose the new training data sources – Recital (109).

The scope of the public disclosure is to be decided. Recital (107) says:

While taking into due account the need to protect trade secrets and confidential business information, this summary should be generally comprehensive in its scope instead of technically detailed to facilitate parties with legitimate interests, including copyright holders, to exercise and enforce their rights under Union law, for example by listing the main data collections or sets that went into training the model, such as large private or public databases or data archives, and by providing a narrative explanation about other data sources used.

Under Article 56(2)(b), the AI Office needs to set out a code of practice covering “the adequate level of detail for the summary about the content used for training” and, under Article 56(9), once the act is in force the code must be ready within 9 months (perhaps by January 2025?).

Protection of IP and trade secrets in AI

The IP and trade secrets of AI providers and their suppliers should be protected. Suppliers need to give AI providers the information needed to comply with the AI Act but without compromising their IP or trade secrets – Recital (88) and Article 25(5). The authorities applying the AI Act should protect IP and trade secrets – Recital (167), Article 52(6), Article 53(7), Article 78 and Annex VII, point 4.5.

IP as a fundamental right

Adverse impact on IP rights is relevant to classifying AI systems as high risk – Recital (48). Time will tell how that plays out – recitals have a significant role in the interpretation of EU law.

Conclusion

Providers of Generative AI must start preparing for compliance (including asking their suppliers for any information needed). In particular, they will need to think long and hard about where and how to train their models to comply with EU copyright law.