Artificial Intelligence Observatory
This column lists relevant news and documents, including legally binding ones, issued at the Italian and European level on Artificial Intelligence.
2025
AI Literacy Q&A Publication
18 Agosto 2025The Commission published detailed Q&A documentation addressing practical implementation of AI literacy requirements.
The guidance clarifies that AI literacy obligations apply to all providers and deployers of AI systems, covering employees, contractors, and service providers interacting with AI systems.
The document emphasizes that organizations must maintain internal records of training initiatives but are not required to issue formal certifications.
General-Purpose AI Code of Practice
11 August 2025Independent experts submitted the GPAI Code of Practice to the Commission, developed through consultation with over 1,000 stakeholders.
This voluntary compliance tool addresses three core areas: transparency obligations, copyright compliance, and safety and security measures.
The Code applies differentially, with transparency and copyright chapters applicable to all GPAI providers, while safety measures target only systemic risk models.
Commission Opinion on GPAI Code
1 August 2025The Commission issued its formal opinion endorsing the GPAI Code of Practice as adequate for demonstrating compliance with AI Act obligations.
This endorsement provides legal recognition of the Code as a valid compliance pathway, though adherence remains voluntary.
Training Content Summary Template and Public Summary
25 July 2025The AI Office released the standardized template for public summaries of GPAI training content, implementing Article 53(1)(d) requirements.
This mandatory disclosure framework requires GPAI providers to publish structured information about training data sources, content types, and collection methodologies.
Then the Commission published an explanatory notice accompanying the training content template, providing detailed instructions for compliance with transparency obligations.
This document clarifies the scope and application of disclosure requirements for GPAI training data.
GPAI Providers Guidelines
18 July 2025The Commission published comprehensive guidelines clarifying the scope of obligations for GPAI model providers.
These guidelines establish technical criteria for determining when models qualify as "general purpose" and specify exemptions for open-source models with certain characteristics.
The document provides pragmatic implementation approaches and clarifies value chain responsibilities.
AI Factories Establishment
19 June 2025The Commission established AI Factories as a new pillar within the EuroHPC Joint Undertaking, providing dedicated supercomputing infrastructure for AI model training.
These facilities enable fast machine learning and training of large GPAI models, offering privileged access to AI startups and the broader innovation community.
The AI Factories serve as one-stop shops for algorithmic development, testing, evaluation, and validation of large-scale AI models.
Study on the deployment of AI in healthcare
June 2025Present day healthcare systems face several complex challenges, including rising demand due to an aging population, increasing prevalence of chronic and complex conditions, rising costs, and shortages in the healthcare workforce.
Artificial intelligence (AI) has the potential to address some of these by improving operational efficiency, reducing administrative burdens, and enhancing diagnosis and treatment pathways.
Despite the promise and availability AI-based tools in the market, their deployment in clinical practice is slow. Using a mixed methods approach, entailing a literature review and consultation activities, the study identifies a range of challenges to AI deployment in healthcare, spanning technological and data-related issues, legal and regulatory complexities, organisational and business challenges, and social and cultural barriers. It also highlights successful strategies (accelerators) employed by hospitals globally to overcome these common obstacles, offering valuable inspiration in the broader European Union (EU) context.
AI Literacy Repository Launch
16 April 2025The Commission published a living repository of AI literacy practices adopted by AI Pact companies, providing practical examples of how organizations ensure sufficient AI literacy within their workforce.
This resource compilation serves as a best practices guide for companies implementing AI literacy programs.
Living guidelines on the responsible use of generative ai in research
15 April 2025This document presents the Living Guidelines on the Responsible Use of Generative AI in Research, developed within the European Research Area Forum under the guidance of the European Commission.
Its objective is to provide researchers, research institutions, and funding bodies with clear, practical recommendations to ensure that generative AI is used in a manner that upholds integrity, transparency, accountability, and respect for ethical and legal standards.
AI Continent Action Plan
9 April 2025The Commission launched the AI Continent Action Plan with the objective of making Europe a global leader in artificial intelligence.
The plan focuses on five key pillars: strengthening AI infrastructure through AI Factories and Gigafactories, creating Data Labs for high-quality data access, launching the Apply AI Strategy for sectoral adoption, developing AI talent through international recruitment initiatives, and establishing the AI Skills Academy.
AI Literacy Webinar and Guidance
20 February 2025The AI Office conducted a webinar explaining AI literacy obligations under Article 4 of the AI Act, providing practical guidance for implementation. The session emphasized that there is no one-size-fits-all approach to AI literacy, requiring organizations to tailor training programs based on technical knowledge, usage context, and affected persons.
Guidelines on AI System Definition
6 February 2025The Commission released detailed guidelines clarifying the seven-element definition of "AI system" under Article 3(1) of the AI Act.
These guidelines assist providers in determining whether software systems constitute AI systems subject to regulatory requirements, addressing the machine-based nature, autonomy levels, adaptiveness, objectives, inference capabilities, output generation, and environmental influence of AI systems.
Guidelines on Prohibited AI Practices
4 February 2025The Commission published comprehensive 135-page guidelines interpreting Article 5 of the AI Act, which establishes absolute prohibitions on certain AI practices.
These guidelines address eight categories of prohibited practices, including manipulative AI techniques, exploitative systems targeting vulnerable groups, social scoring, and real-time biometric identification systems.
The document provides detailed legal explanations and practical examples to ensure consistent application across Member States.
AI Office Institutional Framework
24 January 2025The Commission formally established the European AI Office within its organizational structure, creating the institutional foundation for AI governance at the European level. The AI Office serves as the center of AI expertise and forms the foundation for a unified European AI governance system, tasked with supervising implementation and enforcement of the AI Act.
AI Startup and Innovation Communication
24 January 2025The Commission adopted a strategic communication outlining key activities to support AI innovation, including talent pool strengthening through education and training initiatives, venture capital support, and acceleration of Common European Data Spaces development.
AI Innovation Package Launch
24 January 2025The European Commission launched the AI Innovation Package as a comprehensive support mechanism for European AI startups and SMEs.
This €4 billion investment framework through 2027 focuses on developing trustworthy AI that aligns with EU values and regulatory requirements.
The package includes financial support through Horizon Europe and Digital Europe programmes, with dedicated funding for generative AI development.
2024
Study on Public Sector AI Deployment
4 September 2024Special attention is given to public sector AI deployment, with specific guidance on prohibited practices in law enforcement contexts and transparency requirements for government AI systems.
The framework establishes heightened obligations for public entities deploying AI systems.