Individual safeguards in the era of AI: Fair algorithms, fair regulation, fair procedures

The need for individual safeguards in the face of artificial intelligence (AI) may seem uncontroversial: since AI technologies present risks of violating fundamental rights, conventional legal wisdom would require that, to prevent or mitigate those risks, new safeguards should be created and existing ones, reinforced.Fundamental rights protection has certainly been at the heart of global regulatory and academic concerns regarding AI technologies.In Europe, the original regulatory roadmap was drawn by the European Commission's High Level Expert Group (HLEG).With human dignity as cornerstone of the human-centric approach to AI regulation, the HLEG derived the principles of AI ethics from the fundamental rights enshrined in the EU Treaties, as specific expressions of the EU's foundational values like dignity, freedoms, equality, solidarity and justice.Based on this dignitarian and human-centric premise, the HLEG seminally selected four principles (respect for human autonomy, prevention of harm, fairness and explicability), laying down the normative framework within which the AI Act (AIA) 1 eventually took shape.Following the HLEG, the concept of "safeguard" appears to be consubstantial, if not synonymous with that of "right." This alignment is understandable: rights establish subjective entitlements that provide individuals with specific benefits while imposing obligations to guarantee those benefits on public and private actors (MacCormick, 2008, 110).It, therefore, seems reasonable to assume that protecting individual rights and freedoms requires the (legal) recognition of subjective entitlements.However, as illustrated by the AIA -an instrument focused on safety and risk-prevention -as well as by the contributions to this special issue, rights and freedoms can also be protected through alternative mechanisms, such as safety standards.With this in mind, when we chose the theme for this issue, we understood the notion of "individual safeguard" to mean both a regulatory goal (a framework seeking to safeguard individual rights) and an entitlement (a right to claim a specific type of safeguard or protection).In this context, the compelling question is: how are individual safeguards and fairness connected?Once again, the debate might appear to be uncontroversial.Any reasonable person would no doubt agree that individual safeguards (associated with, say, fundamental rights protection) contribute to a broader sense of fairness.Be that as it may, the safeguards/fairness interrelationship has remained relatively unexplored in AI scholarship.We can point to the digital constitutionalism strand which, indeed, places its focus on fundamental rights protection, but seldom through the fairness lens (De Gregorio, 2022).Building on these observations, our overview of the articles featured in this issue will be organised around two key ideas: (1) fairness requiring legal mechanisms to enhance individual safeguards and (2) individual safeguards designed to promote fair outcomes.