oBSERVATORY

Title: EU Commission Press release "Over a hundred companies sign EU AI Pact pledges to drive trustworthy and safe AI development"
Abstract: On 25 September 2024, the European Commission announced that over 100 companies have signed voluntary pledges under the EU AI Pact, a framework encouraging responsible AI development ahead of the AI Act's full application. The signatories, which include multinational corporations and European SMEs, have committed to actions such as developing AI governance strategies, identifying high-risk AI systems, and promoting AI literacy. Additional commitments include ensuring human oversight and transparency in AI-generated content. The initiative is part of broader EU efforts to boost innovation in AI, with the "AI Factories" initiative supporting the development of AI in various sectors.

Subproject area: Legal Theory
Content: AI
Typology: Press Release
Territorial scope: Supranational

Title: De Lungo, Le prospettive dell’AI generativa nell’esercizio delle funzioni parlamentari di controllo e indirizzo (Federalismi)
Abstract: This paper explores the potential impact of generative AI on representative assemblies, focusing on its ability to enhance the control and direction functions of parliaments. By improving data collection and analysis, generative AI could help alleviate information asymmetries between parliaments, the executive, and technical-bureaucratic bodies. The study also examines how AI may contribute to evidence-based policymaking, regulatory quality, and public policy evaluation, offering insights into the integration of AI in parliamentary procedures, particularly through recent institutional developments such as the NRRP and reforms in European economic governance.

Subproject area: Legal Theory
Content: AI
Typology: Paper

Title: UN Final Report on Governing AI for Humanity
Abstract: This final report from the High-level Advisory Body on Artificial Intelligence examines the need for global AI governance, identifying key gaps in representation, coordination, and implementation. It advocates for enhanced global cooperation through shared principles, standards, and benefits, and explores potential institutional models, including the creation of an international AI agency. The report emphasizes the importance of governance to harness AI for humanity, mitigate risks, and address challenges. It concludes with a call to action for coordinated global efforts in AI governance, supported by international panels, policy dialogues, and capacity-building initiatives.

Subproject area: Legal Theory
Content: AI
Typology: Report
Territorial scope: International

Title: EUROPOL, “AI and policing. The benefits and challenges of artificial intelligence for law enforcement”
Abstract: This report provides law enforcement professionals with a detailed guide on the responsible and compliant use of AI in daily operations. It covers various AI applications, including data analytics, OSINT, natural language processing, and digital forensics, while also addressing ethical concerns such as bias, privacy, and accountability. The report highlights the implications of the EU AI Act for law enforcement, focusing on balancing innovation with ethical compliance. It offers strategies for mitigating risks and fostering public trust, while also forecasting future AI advancements in the law enforcement sector.

Subproject area: Legal Theory
Content: AI, NLP, Big data
Typology: Report
Territorial scope: Supranational

Title: South Africa National AI Policy Framework
Abstract: The National AI Policy Framework for South Africa outlines a strategic plan to leverage AI technologies for economic growth and societal well-being. Prioritizing ethical development, the framework emphasizes fairness, accountability, transparency, and inclusivity to foster trust and mitigate risks like bias and discrimination. Key pillars include robust data governance, infrastructure improvements, and investments in research and innovation. By promoting human-centric AI, the framework seeks to improve quality of life and position South Africa as a leader in AI innovation, addressing challenges in a sustainable and holistic way.

Subproject area: Legal Theory
Content: AI
Typology: Paper
Territorial scope: National

Title: California Bill on Generative artificial intelligence: training data transparency
Abstract: This piece of legislation mandates transparency in the training data of generative AI systems. By January 1, 2026, developers of generative AI systems or services available to Californians must publicly disclose key information about the datasets used for training these systems. This includes a high-level summary of the data. The bill, which builds on existing state requirements for automated decision systems, aims to ensure transparency and accountability in AI development.

Subproject area: Legal Theory
Content: AI
Typology: Legislation
Territorial scope: Sub-national

Title: OECD “Governing with Artificial Intelligence: are Governments Ready?”
Abstract: This policy document highlights the major trends and challenges related to the development, application, and integration of AI within the public sector. Initially, it examines the possible advantages and distinct risks linked to AI implementation in public services. Next, it explores ways in which AI can enhance productivity, responsiveness, and accountability in this sector. Finally, it offers a summary of critical policy matters and shares examples of how various countries within the OECD are tackling these issues.

Subproject area: Legal Theory
Content: AI
Typology: Paper
Territorial scope: International

Title: European Data Protection Supervisor “Generative AI and the EUDPR. First EDPS Orientation for ensuring data protection compliance when using Generative AI systems”
Abstract: On June 3, the EDPS released guidelines concerning generative Artificial Intelligence and personal data for EU institutions, bodies, offices, and agencies (EUIs). These guidelines are designed to assist EUIs in meeting the data protection requirements outlined in Regulation (EU) 2018/1725 while utilising or developing generative AI tools.

Subproject area: Legal Theory
Content: AI, Data Protection
Typology: Soft law
Territorial scope: Supranational

Title: United Nations Human Rights Office of the High Commissioner, “Taxonomy of Human Rights Risks Connected to Generative AI”
Abstract: This document examines the human rights risks associated with the development, implementation, and use of generative AI technology. It aims to illustrate how the most serious harms caused by generative AI are directly related to impacts on internationally recognised human rights.

Subproject area: Legal Theory
Content: AI
Typology: Paper
Territorial scope: International

Title: NIST, AISI “The United States Artificial Intelligence Safety Institute: Vision, Mission, and Strategic Goals”
Abstract: The Strategic Vision document by NIST AISI (Artificial Intelligence Safety Institute) describes the AISI’s philosophy, mission, and strategic goals: advance the science of AI safety; articulate, demonstrate, and disseminate the practices of AI safety; support institutions, communities and coordination around AI safety. To achieve these goals, the AISI will work closely with diverse AI industry, civil society members, and international partners. Its plan is to conduct testing of advanced models and systems to assess potential and emerging risks, develop guidelines on evaluations and risk mitigations, and perform and coordinate technical research. 

Subproject area: Legal theory
Content: AI
Typology: Report
Territorial scope: National

Title: Council of Europe “Council of Europe Framework Convention on Artificial Intelligence and Human Rights. Democracy and the Rule of Law”
Abstract: This is the first international legally binding treaty designed to uphold human rights, the rule of law, and democratic standards in the use of artificial intelligence (AI) systems. Open to non-European countries as well, the treaty establishes a legal framework that encompasses the entire lifecycle of AI systems and addresses the associated risks while encouraging responsible innovation. The convention employs a risk-based approach to the design, development, use, and decommissioning of AI systems, necessitating a thorough evaluation of any potential adverse effects of AI usage.

Subproject area: Legal theory
Content: AI
Typology: Legislation
Territorial scope: International

Title: Generative AI framework for HM Government
Abstract: This framework outlines principles and practical steps for the safe and responsible use of generative AI in HM Government. It includes ten core principles covering lawful and ethical use, security, human control, lifecycle management, tool selection, openness, collaboration, skill acquisition, and policy alignment. The framework delves into the definition of generative AI, its applications, and limitations, while providing guidance on building generative AI solutions, including goal setting, team formation, support structure creation, procurement, and solution development. Legal, ethical, privacy, security, and governance considerations are explored in depth, emphasizing the importance of expert collaboration and alignment with organizational policies.

Subproject area: Technical advanced solutions
Content: AI, NLP, Data Protection
Typology: Report
Territorial scope: National

Title: Camera’s Report on “Utilizzare l’intelligenza artificiale a supporto del lavoro parlamentare”
Abstract: This report presents the outcomes of a study conducted by the Comitato di vigilanza sull’attività di documentazione of the Camera dei deputati concerning the role of artificial intelligence in parliaments. It discusses the significant potential of AI in enhancing societal sectors while acknowledging associated challenges, thus emphasizing the imperative for democratic principles in AI application within parliamentary contexts and beyond. It also provides insights into the current utilization of artificial intelligence by parliaments worldwide. Finally, it outlines guiding principles for the integration of AI in parliamentary processes.

Subproject area: Technical advanced solutions
Content: AI, NLP
Typology: Report
Territorial scope: National

Title: House of Lords Report on “Large language models and generative AI”
Abstract: The report highlights the transformative potential of Large Language Models (LLMs) and the global race to dominate this field. It emphasizes the need for proactive measures to harness economic benefits while mitigating risks to public safety, societal values, and economic competitiveness. Recommendations include preparing for international competition, guarding against regulatory capture, and balancing safety concerns with commercial opportunities. Immediate actions are advised to address security risks and even prepare for potential catastrophic outcomes. Empowering regulators and developing proportionate AI regulation tailored to the UK’s context are also emphasized.

Subproject area: Technical advanced solutions
Content: AI, NLP
Typology: Report
Territorial scope: National

Title: EU Parliament Press release on Artificial Intelligence Act
Abstract: A landmark agreement on the Artificial Intelligence Act has been confirmed by the EU Parliament. Key provisions include safeguards for general-purpose artificial intelligence, limitations on the use of biometric identification systems by law enforcement, bans on social scoring and AI manipulation of user vulnerabilities, and granting consumers the right to launch complaints and receive meaningful explanations. The text awaits a formal adoption in an upcoming Parliament plenary session and final Council endorsement.

Subproject area: Legal theory
Content: AI
Typology: Press Release
Territorial scope: Supranational

Title: EU Commission White Paper on “How to master Europe's digital infrastructure needs?”
Abstract: This White Paper identifies challenges and discusses possible scenarios for public policy actions, such as a possible future Digital Networks Act, that aim to incentivise building the digital networks of the future, master the transition to new technologies and business models, meet future connectivity needs of all end-users, underpin competitiveness of EU economy and ensure secure and resilient infrastructures. A public consultation on the ideas put forward in the Paper is open for Member States, civil society, and academics until June 2024.

Subproject area: Technical advanced solutions
Content: Cybersecurity, Digital assets
Typology: Paper
Territorial scope: Supranational

Title: US NIST Report on “Adversarial Machine Learning. A Taxonomy and Terminology of Attacks and Mitigations”
Abstract: This NIST Trustworthy and Responsible AI report develops a taxonomy of concepts and defines terminology in the field of adversarial machine learning (AML). The taxonomy is built on surveying the AML literature and is arranged in a conceptual hierarchy that includes key types of ML methods and lifecycle stages of attack, attacker goals and objectives, and attacker capabilities and knowledge of the learning process. The report also provides corresponding methods for mitigating and managing the consequences of attacks and points out relevant open challenges to take into account in the lifecycle of AI systems. Taken together, the taxonomy and terminology are meant to inform other standards and future practice guides for assessing and managing the security of AI systems, by establishing a common language and understanding of the rapidly developing AML landscape.

Subproject area: Technical advanced solutions
Content: Cybersecurity, AI, NP
Typology: Report
Territorial scope: National

Title: EU Commission Decision establishing the European Intelligence Office (C/2024/1459)
Abstract: The Decision establishes a European Artificial Intelligence Office within the Commission as part of the administrative structure of the Directorate-General for Communication Networks, Content and Technology to perform tasks necessary to implement and enforce the forthcoming Regulation on artificial intelligence and contribute to fostering EU policies on AI.

Subproject area: Legal theory
Content: AI
Typology: Legislation
Territorial scope: Supranational

Title: Artificial Intelligence in the European Commission (AI@EC) C(2024) 380 final
Abstract: In this Communication the Commission develops internal operational guidelines that give staff clear and pragmatic guidance how to put AI systems in operation, classifies and assesses AI systems in use – or planned to be used – based on a risk-based approach, sets out AI uses that are considered incompatible with European values or that represent a threat to the security, safety, health, and fundamental rights of people, and puts in place organisational structures to fulfil the obligations of the Commission in relation to AI, also for the forthcoming AI Act.

Subproject area: Legal theory
Content: AI
Typology: Soft-law
Territorial scope: Supranational

Title: UN-AI Advisory Body Interim Report on “Governing AI for Humanity”
Abstract: The report proposes preliminary recommendations for AI governance categorized into guiding principles and institutional functions. Guiding principles emphasize inclusivity, public interest, alignment with data governance and promotion of data commons, multi-stakeholder collaboration, and adherence to international commitments. Institutional functions focus on assessing AI’s future implications, reinforcing governance interoperability, developing standards and risk management frameworks, facilitating international cooperation, promoting talent development and the SDGs, and ensuring compliance and accountability through binding norms.

Subproject area: Legal theory
Content: AI
Typology: Report
Territorial scope: International

Title: COE’s Draft Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law
Abstract: This document presents the outcomes of the second reading of the Draft Framework Convention on Artificial Intelligence, Human Rights, Democracy, and the Rule of Law by the Committee on Artificial Intelligence. The Convention establishes general principles aimed at ensuring the protection of human rights, the integrity of democratic processes, and the respect for the rule of law. These principles include safeguarding human dignity and autonomy, promoting transparency and oversight, ensuring accountability, fostering equality, protecting privacy, preserving health, promoting reliability, and encouraging safe innovation. Additionally, the Convention outlines obligations to implement procedural safeguards and mechanisms for risk assessment and mitigation.

Subproject area: Legal theory
Content: AI
Typology: Bill
Territorial scope: International

Title: D.M. Katz, Natural Language Processing in the Legal Domain
Abstract: The paper offers an overview of the current state of NLP & Law, focusing on recent advancements in both technical and substantive aspects. Analyzing a corpus of over six hundred papers published in the last decade, the study identifies increasing trends in the number of papers, tasks undertaken, and languages covered. Researchers are employing more sophisticated methodologies, aligning Legal NLP with broader scientific standards. While these developments are promising, many unanswered questions remain in both academic and commercial spheres.

Subproject area: Technical advanced solutions
Content: NLP
Typology: Paper
Territorial scope: International

Title: R. Bommassani et al., The Foundation Model Transparency Index
Abstract: The paper highlights the rise of foundation models in AI applications alongside a decline in transparency. It introduces the 2023 Foundation Model Transparency Index, which evaluates major developers' transparency across 100 indicators. Gaps in transparency, especially regarding downstream impacts, are evident. This index aims to drive improvements in governance through industry standards and regulation.

Subproject area: Legal theory
Content: AI
Typology: Paper
Territorial scope: National

Title: CNIL guidelines for a use of AI respectful of personal data
Abstract: The article outlines CNIL guidelines for aligning AI innovation with GDPR principles, promoting technological progress while safeguarding privacy. These guidelines detail GDPR's application to AI, stressing compatibility and setting conditions for data handling. Integrating privacy measures fosters trust in ethical AI development, bolstering confidence in emerging technologies.

Subproject area: Legal theory
Content: AI
Typology: Press releases
Territorial scope: National

Title: EU Commission’s report on “Cybersecurity of Artificial Intelligence in the AI Act”
Abstract: The European Commission’s proposal for the AI Act represents a significant milestone in the regulation of Artificial Intelligence (AI). This report focuses on the cybersecurity requirement for high-risk AI systems, as set out in Article 15 of the regulation. It presents a high-level analysis in the context of the rapidly evolving AI landscape, and provides a set of key guiding principles to achieve compliance with the AI Act.

Subproject area: Legal theory
Content: AI
Typology: Report
Territorial scope: Supranational

Title: EU Commission’s draft “Proposal for standard contractual clauses for the procurement of Artificial Intelligence (AI) by public organisations”
Abstract: The EU Commission’s report introduces standard contractual clauses for public organizations procuring AI Systems. These clauses align with the proposed AI Act, but their adoption is voluntary and adaptable on a case-by-case basis. They focus on high-risk AI systems, but a simplified version is available for non-high-risk AI. These clauses specifically address AI-related matters under the AI Act and are meant to be appended to existing agreements, excluding other contractual obligations.

Subproject area: Legal theory
Content: AI
Typology: Report
Territorial scope: Supranational

Title: WEF paper on “Exploring the Industrial Metaverse”
Abstract: The briefing paper by WEF in collaboration with the University of Cambridge outlines a roadmap for industry leaders navigating the nascent industrial metaverse. It highlights four key dimensions: market assessment, use case experimentation, technology infrastructure, and resource management. As the industrial metaverse is set to expand, understanding its multidimensional impact is crucial for organizations and society.

Subproject area: Legal theory
Content: Metaverse
Typology: Report
Territorial scope: International

Title: EU Parliament’s study on “The emergence of non-personal data markets”
Abstract: The European Union has positioned itself uniquely by promoting the free flow of non-personal data while maintaining stringent data privacy and security measures. Nonetheless, ensuring fair conditions for data sharing remains essential. This study, presented by the Policy Department for Economic, Scientific, and Quality of Life Policies at the request of the Committee on Industry, Research, and Energy (ITRE), delves into the emergence of non-personal data markets. It examines the existing EU legal framework, identifying ongoing challenges and suggesting implementation strategies.

Subproject area: Legal theory
Content: Big data
Typology: Report
Territorial scope: Supranational

Title: Datatilsynet’s report on “Using artificial intelligence in the public sector”
Abstract: The Danish Data Protection Agency (Datatilsynet) study examines AI usage in the public sector, emphasizing the need for legal measures to address challenges. It explores AI's impacts on public administration and investigates its prevalence among public authorities, focusing on legal considerations and data protection compliance. Through a questionnaire survey, the study provides insights into AI adoption and its implications for data protection. Despite limitations, it sheds light on AI's current use in the public sector, stressing the importance of regulatory oversight for responsible deployment.

Subproject area: Legal theory
Content: AI
Typology: Report
Territorial scope: National

Title: D. Liga, The Interplay Between Lawfulness and Explainability in the Automated Decisionmaking of the EU Administration
Abstract: This work has two main goals, on the one side it explores the nature of explainability in the attempt to clarify the ambiguous use of this concept and how eXplainable AI (XAI) methods fit into this concept. On the other side, the work describes the legal framework which currently regulates explainability of automated decisions in the context of the European administration, showing to what extent a selection of famous XAI methods meets the requirements of such legal framework.

Subproject area: Legal theory
Content: AI
Typology: Paper
Territorial scope: Supranational

Title: UNIDROIT Principles on Digital Assets and Private Law
Abstract: The UNIDROIT Principles on Digital Assets and Private Law aim to regulate transactions involving digital assets, reducing legal uncertainties, and promoting efficiency. Neutral regarding technology and jurisdiction, they apply across different legal systems. They cover specific aspects of digital transactions while excluding other areas of private law. They address transfers, security, custody, and legal procedures, providing guidelines for a gradual transition.

Subproject area: Legal theory
Content: Digital assets
Typology: Soft-law
Territorial scope: International

Title: German Federal Office for Information Security on “Generative AI Models”
Abstract: This document is the second edition of a report on the opportunities and risks of Generative AI, particularly focusing on LLMs, for industries and authorities. The publication aims to cultivate a fundamental understanding of security considerations for these models and promote their safe integration into existing workflows. To achieve this, it not only highlights opportunities but also emphasizes the most significant current dangers and resulting risks during the planning, development, and operational phases, as well as the utilization of generative AI models. Additionally, it suggests possible countermeasures throughout the entire lifecycle of the models.

Subproject area: Technical advanced solutions
Content: AI, NLP, Data Protection
Typology: Report
Territorial scope: National

Title: EPRS At a glance on “Protecting children in virtual worlds (the metaverse)”

Abstract: This report, authored by the European Parliamentary Research Service, delves into the opportunities and challenges presented by virtual worlds for children. It provides insights into the actions undertaken by both the European Union and the private sector to effectively manage these complexities. The document sheds light on the evolving landscape of virtual environments and their applications and offers recommendations for ensuring the safety and well-being of young users.

Subproject area: Legal theory, Technical advanced solutions
Content: Metaverse, Data protection
Typology: Report
Territorial scope: Supranational

Title: EPRS At a glance on “Artificial intelligence and cybersecurity”
Abstract: This report, authored by the European Parliamentary Research Service, analyzes the intersections of AI and Cybersecurity and tackles the most recent issues related to the cybersecurity of AI, the use of AI to support cybersecurity, and the use of AI for malicious purposes. For all these dimensions, the report highlights current standards and use cases.

Subproject area: Legal theory, Technical advanced solutions
Content: AI, Cybersecurity
Typology: Report
Territorial scope: Supranational

Title: Italian strategy for AI 2024-2026
Abstract: The document delineates Italy's AI strategy for the forthcoming years. This strategy encompasses initiatives aimed at fostering a regulatory framework aligned with ethical standards and social accountability, while taking into account various interconnected factors such as privacy, security, gender equality, environmental concerns, and sustainability. These initiatives span across four key sectors: Research, Public Administration, Businesses, and Education, and are reinforced by efforts to bolster innovation-enabling infrastructure and establish a framework for the execution, coordination, and oversight of activities.

Subproject area: Legal theory, Technical advanced solutions
Content: AI
Typology: Report
Territorial scope: National

Title: US NIST “Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile”

Abstract: The publication by the US Department of Commerce serves as a companion resource to the AI Risk Management Framework (AI RMF), aligning with President Biden's Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence. It identifies a group of risks unique to or exacerbated by Generative AI (GAI), as acknowledged by the GAI Public Working Group (PWG). These risks include CBRN Information, Confabulation, Dangerous or Violent Recommendations, Data Privacy, Environmental concerns, Human-AI Configuration, Information Integrity, Information Security, Intellectual Property, Obscene, Degrading, and/or Abusive Content, Toxicity, Bias, and Homogenization, and Value Chain and Component Integration. After analysing them, the document offers a series of actions to assist organizations in governing, mapping, measuring, and managing these risks.

Subproject area: Technical advanced solutions
Content: Cybersecurity, Digital assets
Typology: Paper
Territorial scope: National

Title: United Nations “Seizing the opportunities of safe, secure and trustworthy artificial intelligence systems for sustainable development”

Abstract: In this resolution, the UN General Assembly calls for responsible AI development, stressing human rights and international cooperation to bridge digital divides and ensure equitable access to AI benefits while addressing misuse risks.

Subproject area: Legal theory
Content: AI
Typology: Soft-law
Territorial scope: International

Title: Australia’s National Science Agency “Artificial Intelligence foundation models”

Abstract: The report was prepared by Australia’s National Science Agency to inform industry, government and community organisations about the strategic implications of AI foundation models for Australia.

Subproject area: Technical advanced solutions
Content: AI
Typology: Report
Territorial scope: Supranational

Title: Commission de l’intelligence artificielle, “IA: notre ambition pour la France”

Abstract: The text by the AI Commission of the French Government proposes strategic actions for France in the AI revolution, focusing on education, innovation investment, computational infrastructure, data access, research freedom, and global governance to position France as an AI leader and ensure societal benefits through collective commitment.

Subproject area: Legal theory, technical advanced solutions
Content: AI
Typology: Report
Territorial scope: Supranational

Title: European Parliament “What Think Tanks are Thinking"

Abstract: This note by the EU Parliament offers links to recent reports and commentaries from some major international think tanks and research institutes on artificial intelligence.

Subproject area: Legal theory, technical advanced solutions
Content: AI
Typology: Report
Territorial scope: Supranational

Title: State of California, Department of Technology, Office of Information Security “Generative Artificial Intelligence Risk Assessment”

Abstract: To proactively address potential threats to state-owned information assets, privacy, andthe welfare of California's citizens, the Statewide Information Management Manual(SIMM) 5305-F, Generative Artificial Intelligence (GenAI) Risk Assessment introduces arisk assessment methodology that will aid state entities in evaluating the risksassociated with GenAI systems. This SIMM is to ensure alignment with Executive Order (EO) N-12-23, The White HouseBlueprint for an GenAI Bill of Rights, The White House EO on Safe, Secure, andTrustworthy Artificial Intelligence, and the National Institute of Standards andTechnology (NIST) Artificial Intelligence Risk Management Framework.

Subproject area: Technical advanced solutions
Content: AI
Typology: Report
Territorial scope: Supranational