The video at the end of this article contains a training on Artificial Intelligence (AI) in Horizon Europe, given by Cristina Contero Almagro (Aphaia). Over the past five years, Europe has prioritized AI as a means to enhance its competitive edge. This training provides an overview of key milestones in AI regulation, including the ethics guidelines for trustworthy AI, the Assessment List for justworthy AI, and the proposed Artificial Intelligence Act. The focus lies on what you can do within the frame of self-assessment, and is especially relevant for AI-heavy projects.

You can find below the basic information (and links to more information) of the most important documents.

Ethics Guidelines for Trustworthy AI

The foundation of AI regulation began with the Ethics Guidelines for Trustworthy AI in 2018. The guidelines defined trustworthy AI based on three components: legality, ethical adherence, and robustness. To achieve trustworthy AI, compliance with laws, ethical principles, and robust technical and societal frameworks is essential to prevent unintended harm and enhance societal acceptance of AI technologies.

Assessment List for Trustworthy AI

Following the ethics guidelines, the Assessment List for Trustworthy AI was published in late 2019, serving as an operational tool. It outlines specific requirements and questions related to ethical AI, aiding applicants and stakeholders in understanding and assessing their projects' adherence to ethical standards. While not offering definitions, it prompts critical self-reflection to ensure ethical considerations are adequately addressed.

The Proposed Artificial Intelligence Act

The Artificial Intelligence Act, although not yet applicable, represents a significant step in AI regulation. It categorizes AI systems into four risk levels: acceptable risk, high risk (e.g., critical infrastructures), limited risk (e.g., standard websites), and minimal risk (e.g., video games). Each category corresponds to distinct regulatory requirements, reinforcing transparency and accountability, especially in high-risk applications.

Privacy and Data Governance Requirements

Privacy and data governance are key aspects linked to principles of preventing harm, fairness, and accountability. This requirement emphasizes the critical role of data management and privacy protections. Ensuring compliance with data privacy regulations is vital in the majority of AI projects, highlighting the need for familiarity with these aspects in AI development and implementation.

Understanding the evolving landscape of AI regulation in Horizon Europe is fundamental for successful project applications. The Ethics Guidelines, Assessment List, and proposed Artificial Intelligence Act provide a roadmap for developers and stakeholders to navigate the legal and ethical implications of AI. Addressing privacy and data governance concerns is equally vital to ensure responsible and effective AI deployment within the European context.