BIG DATA, ALGORITHMS AND FOOD SAFETY : a legal and ethical approach to data ownership and data... governance

This book identifies the principles that should be applied when processing Big Data in the context of food safety risk assessments. Food safety is a critical goal in the protection of individuals’ right to health and the flourishing of the food and feed market. Big Data is fostering new applications capable of enhancing the accuracy of food safety risk assessments. An extraordinary amount of information is analysed to detect the existence or predict the likelihood of future risks, also by means of machine learning algorithms. Big Data and novel analysis techniques are topics of growing interest for food safety agencies, including the European Food Safety Authority (EFSA). This wealth of information brings with it both opportunities and risks concerning the extraction of meaningful inferences from data. However, conflicting interests and tensions among the parties involved are hindering efforts to find shared methods for steering the processing of Big Data in a sound, transparent and trustworthy way. While consumers call for more transparency, food business operators tend to be reluctant to share informational assets. This has resulted in a considerable lack of trust in the EU food safety system. A recent legislative reform, supported by new legal cases, aims to restore confidence in the risk analysis system by reshaping the meaning of data ownership in this domain. While this regulatory approach is being established, breakthrough analytics techniques are encouraging thinking about the next steps in managing food safety data in the age of machine learning. The book focuses on two core topics – data ownership and data governance – by evaluating how the regulatory framework addresses the challenges raised by Big Data and its analysis in an applied, significant, and overlooked domain. To do so, it adopts an interdisciplinary approach that considers both the technological advances and the policy tools adopted in the European Union, while also assuming an ethical perspective when exploring potential solutions. The conclusion puts forward a proposal: an ethical blueprint for identifying the principles – Security, Accountability, Fairness, Explainability, Transparency and Privacy – to be observed when processing Big Data for food safety purposes, including by means of machine learning. Possible implementations are then discussed, also in connection with two recent legislative proposals, namely the Data Governance Act and the Artificial Intelligence Act. Acknowledgements Contents Acronyms 1 Food, Big Data, Artificial Intelligence 1.1 Emerging Trends in Food Technology 1.1.1 Big Data, the Food Industry and the UN Sustainable Development Goals 1.1.2 Sensors and Traceability in Agriculture 4.0 1.1.3 The in silico Age of Food Safety Risk Assessment: From Food Tasters to Machine Learning 1.2 Hard Trade-Offs in Data Ownership and Data Governance: Mapping the Debate 1.2.1 Property Rights, Confidentiality and Transparency 1.2.2 Information Privacy, Food Consumption and Data Transmissions 1.2.3 Big Data and Competition in the Food Industry 1.3 Novel Trends in Food-Related Data Processing 1.3.1 'Data Ethics' and `Soft Ethics' 1.3.2 The Untapped Relevance of Food Consumption Information 1.3.3 Big Data, AI, the Role of Principles in the Forthcoming Regulation 1.4 Chapter Synopsis References 2 Data Ownership in Food-Related Information 2.1 Traditional and Emerging Approaches to Data Ownership 2.1.1 Ownership Models and Data 2.1.2 Confidentiality and Digital Commons 2.1.3 Transparency, Truthfulness, Trust 2.2 Regulatory and Policy Trends in Openness and Transparency 2.2.1 The Aarhus Convention and the Right to Access Environmental Information 2.2.2 Transparency of Food Safety Information in the EU Following the 2021 Reform (Reg. 2019/1381) 2.2.2.1 What Information? The EFSA Data Warehouse 2.2.2.2 General Scope of the GFLR: Legal Basis, Goals and the Precautionary Principle 2.2.2.3 EFSA, Data Collection and Storage 2.2.2.4 EFSA Data Transparency Obligations 2.2.2.5 EFSA Confidentiality Management 2.2.2.6 Regulation 2019/1381 2.2.2.7 New Articles 32a `Pre-submission Advice', 32b `Notification of Studies', 32c `Consultation of Third Parties', 32d `Verification Studies' 2.2.2.8 Amended Art. 38 `Transparency' 2.2.2.9 Amended Art. 39 `Confidentiality' 2.2.2.10 New Articles 39a `Confidentiality Request', 39b `Decision on Confidentiality', 39c `Review of Confidentiality', Art. 39d `Obligations with Regard to Confidentiality' 2.2.2.11 New Art. 39e: `Protection of Personal Data' 2.2.2.12 New Art. 39f `Standard Data Formats', Art. 39g `Information Systems' 2.2.3 Recent Case Law of the European Court of Justice: A Short Commentary 2.2.3.1 Greenpeace and PAN Europe on the Concept of Environmental Information and on the Balance of Public and Commercial Interests 2.2.3.2 Tweedale and Hautala on the Overriding Public Interest on the Disclosure of Confidential Dossiers 2.2.3.3 Arysta LifeScience Netherlands BV v EFSA on Confidentiality Claims Decisions 2.3 A New Paradigm of Data Ownership in Food-Related Information 2.3.1 Open Questions in Food Safety Data Ownership 2.3.2 Towards a New Concept of Environmental Information 2.3.3 Trust-Oriented Approaches to Data Ownership 2.3.4 Dynamic Ownership and Shared Benefits 2.4 Chapter Synopsis References 3 Food Consumption Data Protection 3.1 Food Consumption, Surveys and Mobile Applications 3.1.1 Food Safety Dietary Intake Surveys 3.1.2 Food Diaries, Data Mash-Up, Mixed Datasets 3.1.3 Food Consumption Data, Derived and Inferred Information 3.2 Dietary Information Processing and Data Protection Between Law and Ethics 3.2.1 Quasi-Personal and Quasi-Sensitive Data 3.2.2 Food Consumption and Limitations in the Notion of Personal Data 3.2.3 Group Privacy and Dietary Preferences 3.3 ``You Are What You Eat'': Food and Personal Identity 3.3.1 The Datafication of Food Preferences 3.3.2 From Food Preferences to Personal Identity 3.3.3 The Next Steps in Dietary Intake Data Protection 3.4 Chapter Synopsis References 4 Current Trends, Machine Learning, and Food Safety Data Governance 4.1 Data and Algorithmic Transparency 4.1.1 The Shift from Deterministic to Stochastic Analysis 4.1.2 Possible Biases in Food Consumption Data Analysis 4.1.3 Opening the Black Box: The Knowability Approach 4.2 Accountability and Redress of Probabilistic Models 4.2.1 Fairness in Risk Assessment 4.2.2 Responsibility, Accountability, Liability 4.2.3 Machine Learning and the Precautionary Principle 4.3 The EU Data Governance Act and its Impact on Food Data Governance 4.3.1 The Proposal for an EU Data Governance Act: An Overview 4.3.2 The Applicability of `Data Exclusivity' Clauses and Competition Issues 4.3.3 Data Altruism, Data Cooperatives, General and Public Interest 4.4 Chapter Synopsis References 5 The P-SAFETY Model: A Unifying Ethical Approach 5.1 Towards an Ethical Blueprint 5.1.1 Trust-Centric Ethics of Artificial Intelligence 5.1.2 Risk-Based and Principle-Based Approaches to AI Regulation 5.1.3 Institutional AI Charters 5.1.3.1 European Commission Communications: ``Artificial Intelligence for Europe'' and ``White Paper On Artificial Intelligence'' 5.1.3.2 German Federal Government: ``AI Strategy'' 5.1.3.3 French Conseil National du Numerique: ``For a Meaningful Artificial Intelligence: Towards a French and European Strategy'' 5.1.3.4 SIGAI, The Special Interest Group of AI: ``Dutch AI Manifesto'' 5.1.3.5 AGID, Agency for Digital Italy: ``White Paper on Artificial Intelligence at the Service of the Citizen'' 5.1.4 Interdisciplinary Ethical Charters 5.1.4.1 Asilomar Conference 5.1.4.2 IEEE Ethically Aligned Design 5.1.4.3 AI4People Initiative 5.1.4.4 EU Commission High Level Expert Group: ``Guidelines for Trustworthy AI'' 5.2 The P-SAFETY Blueprint for Trustworthy Food Safety 5.2.1 Security 5.2.2 Accountability 5.2.3 Fairness 5.2.4 Explainability 5.2.5 Transparency 5.2.6 The Question of Privacy: From SAFETY to P-SAFETY 5.3 Possible Implementations of the P-SAFETY Blueprint 5.3.1 The Middle-Out Approach 5.3.2 Hard Law, Self-Regulation, Codes of Conduct 5.3.3 Case-Law and Judicial Interpretation 5.4 Chapter Synopsis References 6 Conclusion: A Responsible Food Innovation 6.1 Restoring Information Asymmetries 6.2 Food Safety Authorities, the Industry and AI: Joint Efforts Towards Trustworthiness 6.3 The Next Steps in Food Data Governance Reference