Article 15 of the AI Act seeks to contribute to the trustworthiness of AI systems. According to Article 15, trustworthiness means requiring the developing and deployment of AI systems that contain features such as accuracy, robustness, and cybersecurity. For instance, focusing on protecting the data and privacy of citizens against cyberattacks or developing AI systems with accurate data to promote non-harmful outputs. Arguably, the requirements of ‘trustworthy AI’ through the principles outlined in Article 15 of the AI Act can lead to closing accountability gaps and opening legal opportunities to contest the liability of developers and deployers of AI systems.

Human analysts working for threat intelligence leverage tools powered by artificial intelligence to routinely assemble actionable intelligence. Yet, threat intelligence sources and methods often have significant uncertainties and biases. In addition, data sharing might be limited for operational, strategic, or legal reasons. Experts are aware of these limitations but lack formal means to represent and quantify these uncertainties in their daily work. In this chapter, we enunciate the technical, legal, and societal challenges for building explainable AI for threat intelligence. We also discuss ideas for overcoming these challenges.

Accountability is considered a cornerstone of public administration and good governance. This study characterizes the relationship between the Dutch Intelligence and Secret Service (“AIVD”) and citizens (represented by parliament, courts, and oversight boards) as a complex actor-forum relationship. We utilize different accountability principles of public administration found in international and Dutch instruments and academic literature to propose workable principles of accountability for the AIVD. These proposed principles of accountability can be summarized as acting within duty, explainability, necessity, proportionality, reporting and record keeping, redress, and continuous independent oversight. Similarly, there are some conditions to support the workability of accountability principles. These conditions may be characterized as productive actor-forum relationships, cooperation, flexibility, value alignment, and learning and improving opportunities.

The AI Act represents a significant legislative effort by the European Union to govern the use of AI systems according to different risk-related classes, imposing different degrees of compliance obligations to users and providers of AI systems. However, it is often critiqued due to the lack of general public comprehension and effectiveness regarding the classification of AI systems to the corresponding risk classes. To mitigate these shortcomings, we propose a Decision-Tree-based framework aimed at increasing legal compliance and classification clarity. By performing a quantitative evaluation, we show that our framework is especially beneficial to individuals without a legal background, allowing them to enhance the accuracy and speed of AI system classification according to the AI Act. The qualitative study results show that the framework is helpful to all participants, allowing them to justify intuitively made decisions and making the classification process clearer.

The AI Act represents a significant legislative effort by the European Union to govern the use of AI systems according to different risk-related classes, linking varying degrees of compliance obligations to the system's classification. However, it is often critiqued due to the lack of general public comprehension and effectiveness regarding the classification of AI systems to the corresponding risk classes. To mitigate those shortcomings, we propose a Decision-Tree-based framework aimed at increasing robustness, legal compliance and classification clarity with the Regulation. Quantitative evaluation shows that our framework is especially useful to individuals without a legal background, allowing them to improve considerably the accuracy and significantly reduce the time of case classification.

The European Commission has given emphasis to a human-centric approach to allow the deployment of “safe” Artificial Intelligence (AI) systems in our society. However, Article 14 (human oversight obligations) of the EU AI Act Proposal provides little emphasis on human overseers’ responsibilities when tasked to perform meaningful human oversight. This paper tests Article 14 against different challenges of public service organisations such as police departments (street-level bureaucracies) and argues why Article 14 may not be a success in street-level bureaucracies, which could ultimately harm fundamental rights.