The third draft of the General-Purpose AI Code of Practice: Objectives and future perspectives

The rapid advancement of artificial intelligence (AI) technologies has necessitated the development of comprehensive frameworks to ensure their ethical and responsible use. In this context, the European Commission has introduced the third draft of the General-Purpose AI (GPAI) Code of Practice, aiming to guide AI model providers in aligning with the stipulations of the EU AI ActRegulation (EU) 2024/1689). This article delves into the primary objectives of this draft and explores its future implications.

I – Objectives of the third draft of the General-Purpose AI Code of Practice

A – Enhancing transparency

A cornerstone of the third draft is its emphasis on transparency. All providers of general-purpose AI models are mandated to disclose pertinent information about their models, including design specifications, training data sources, and intended applications. This initiative seeks to foster trust among users and stakeholders by ensuring they are well-informed about the functionalities and potential limitations of AI systems. Notably, certain open-source models are exempted from these transparency obligations, reflecting a nuanced approach to diverse AI development paradigms.

B – Addressing copyright concerns

The draft also tackles the intricate issue of copyright in AI development. Providers are required to implement measures that respect intellectual property rights, ensuring that AI models do not infringe upon existing copyrights. This includes establishing mechanisms for rights holders to report potential violations and for providers to address such claims effectively. The draft outlines that providers may refuse to act on complaints deemed “manifestly unfounded or excessive, particularly due to their repetitive nature.”

C – Ensuring safety and security

For AI models identified as posing systemic risks, the draft delineates additional commitments focused on safety and security. Providers of these advanced models are obligated to conduct comprehensive risk assessments, implement robust mitigation strategies, and establish incident reporting protocols. These measures aim to preemptively address potential threats and ensure that AI systems operate within safe and ethical boundaries.

II – Future perspectives of the General-Purpose AI Code of Practice

A – Implementation challenges

As the AI landscape continues to evolve, implementing the GPAI Code of Practice presents several challenges. Providers must navigate the complexities of aligning their operations with the Code’s requirements, which may necessitate significant adjustments in their development and deployment processes. Ensuring compliance without stifling innovation will be a delicate balance to maintain.

B – Global influence and harmonization

The GPAI Code of Practice has the potential to set a global benchmark for AI governance. By establishing comprehensive guidelines, the European Commission aims to influence international standards, promoting harmonization across jurisdictions. This could lead to a more cohesive global approach to AI regulation, benefiting both providers and users worldwide.

Alongside the third draft, the Chairs and Vice-Chairs are also introducing a dedicated executive summary and an interactive website. These resources aim to facilitate stakeholder input, both through written comments and discussions within working groups and specialized workshops. The final version of the Code is expected in May, serving as a compliance framework for general-purpose AI model providers under the AI Act, while integrating cutting-edge best practices.

C – Continuous evolution and adaptation

Recognizing the rapid pace of AI advancements, the Code is designed to be adaptable. It emphasizes the need for continuous evolution, allowing for updates and refinements that reflect technological progress and emerging ethical considerations. This flexibility ensures that the Code remains relevant and effective in guiding AI development responsibly.

Conclusion

The third draft of the General-Purpose AI Code of Practice represents a significant step toward responsible AI governance. By focusing on transparency, copyright respect, and safety, it lays a foundation for ethical AI development. As the cCde progresses toward finalization, its successful implementation will depend on collaborative efforts among stakeholders to address challenges and seize opportunities for global harmonization.

Need expert guidance on AI and intellectual property? Dreyfus Law Firm specializes in intellectual property law, including trademark, copyright, and AI-related legal matters.

We collaborate with a global network of intellectual property attorneys.

Join us on social media !

FAQ