Workshop on AI, Ethics and Morality Workshop on 20 June Ethical and human centric? A protestant view on the AI Act of the European Union Katrin Hatzinger, director of the Brussels Office of the Protestant Church in Germany (EKD)

Whilst recognising the importance of technical innovation and potential benefitsThe EKD-office supported, since the beginning, the risk-based approach, which focuses on the potential harmful effects of AI applications for individuals, societies and the entire creation.

The aim of the AI Act as adopted this year, is to promote the introduction of human-centred and trustworthy AI, while ensuring a high level of protection of health, safety and fundamental rights. EKD is of the opinion that a dialogue platform with equal representation, including theologians, should be set up to accompany the implementation of the legislation to monitor developments and to foster dialogue and exchange within society.

Ladies and gentlemen,

Thank you very much for the invitation and the opportunity to present our work on the EU AI Act in this workshop. My intervention will be divided in three parts: I. A few words on our “church representation” in Brussels, II. EKD´s view on AI and III. our work on the AI act.

1. In dialogue with the European Union: The Brussels Office of the Protestant Church in Germany (EKD)

Allow me to start with a short overview of our work in Brussels in order to set the scene. The Protestant Church in Germany (abbreviated “EKD”) comprises 20 regional protestant churches in Germany. Together, these regional churches represent about 18.5 million registered members from the Lutheran, the Reformed and the United traditions. Even though there is a clear separation between church and  state, the churches in Germany are organised as public entities which has consequences for their organisational structure. EKD´s structure is based on federal principles and there is a tradition of having church representations vis-à-vis state authorities at a federal level in Berlin and at the level of the German states. Due to the increasing importance of European legislation, in 1990, EKD established an office in Brussels as to be represented vis-à-vis the EU institutions.

The director of the Brussels office is traditionally a German lawyer. The team comprises nine permanent staff. The office serves different purposes: Originally it was set up to monitor European legislation and to “defend” the special state/church-relation as codified in the German constitution against European interference. Today, the office is also very committed to advocacy work. It feeds, inter alia, protestant positions on a variety of issues into the political and legal processes on EU level: This comprises topics such as asylum and migration, social, climate, cohesion and youth policies as well as digitalisation or security and defence politics. In order to develop positions on EU legislation we exchange with the colleagues in our Berlin office as well as with colleagues in our headquarters in Hanover. We also rely on EKD publications and working groups in this regard.

Moreover, the office informs church bodies and institutions on current European developments, organises educational trips of church delegations to Brussels, offers a space for public debates and issues a bi-annual newsletter on European topics from a church perspective. The tasks of the office also include networking with other faith-based organisations, political foundations, think tanks and secular NGOs. Finally, we set up a specialised unit working on access to EU funding for projects of protestant churches and related institutions.

II. EKD´s perspective(s) on Artificial Intelligence (AI)

EKD, as an organisation, has not yet developed a common position in form of a memorandum (“Denkschrift”) on the ethics of AI. However, there are single papers and statements, mainly by bishops and protestant academics reflecting on the topic.

Moreover, there is a memorandum dealing with digitalisation in more general way. This memorandum with the title “Freedom digital- Freiheit digital - The ten Commandments in times of the digital transformation” was published in 2021 and points out the relevance of the ten Commandments in times of digitalisation.

The background of the memorandum was to develop protestant ethics in the light of the current digital transition with the aim to open spaces for discourse, providing orientation and enabling people to form their own opinions. The theological key message is the challenge to use the newly gained freedom thanks to the technology in a responsible way towards God and human beings. Thus, technology should be used for a more humane development of society.

The memorandum does not deal with AI as such but touches upon some elements linked to the AI debate like the power of algorithms, autonomous weapon systems and it reflects the notion of trust (relating to the 9th Commandment) when it comes to fake news, deep fakes and conspiracy theories spread by AI.

3. Engaging in the political debate: The position of the EKD-office in Brussels on the AI Act of the European Union

Against this background, me and my colleagues in the EKD-office in Brussels have closely followed the political debates on regulating artificial intelligence in the European Union from an ethical perspective.

1. Consultations, Deliberations, Statements – How the EKD office in Brussels followed the EU debates on the AI Act

It all started back in 2018, when the Commission published a communication titled "Artificial Intelligence for Europe" the European Union's approach to artificial intelligence and robotics. According to this communication, the EU's aim was to secure competitiveness and to shape the development and use of AI while respecting European values. The EU's approach was based on three pillars. These include (1) increasing the budget to strengthen research on AI in Europe; (2) preparing for socio-economic changes brought about by AI to support member states in changing their education and labour policies; and (3) ensuring an appropriate ethical and legal framework to be prepared for such issues.

For the latter, the European Commission appointed a group of experts who presented draft ethical guidelines which could be commented on by the interested public. There was, to our regret, no theological expertise represented in the expert group.

The EKD Brussels Office participated in the consultation on the draft and criticised that it was –given the composition of the expert group - highly influenced by business interests.

The final version of the guidelines were published in April 2019. According to the Guidelines, trustworthy AI should be:

(1) lawful - respecting all applicable laws and regulations

(2) ethical - respecting ethical principles and values

(3) robust - both from a technical perspective while taking into account its social environment

The guidelines list four ethical principles, rooted in fundamental rights, which must be respected in order to ensure that AI systems are developed, deployed and used in a trustworthy manner. These are the principles of:

• Respect for human autonomy
• Prevention of harm
• Fairness and
• Explicability

Against this background, the European Commission published in February 2020 a White paper and a roadmap on AI which we also commented on, arguing inter alia for a regulatory framework.

In April 2021, the European Commission abandoned its soft law approach and presented a draft of the world's first comprehensive legal framework for artificial intelligence: the AI regulation. The AI Regulation is a product regulation, i.e. AI as such is not regulated, but certain AI applications. The Commission proposal laid down harmonized rules for the placing on the market, putting into service and use of AI systems in the European Union. Most of the regulations fall on providers (developers) of high-risk AI systems. Deployers (natural or legal persons deploying an AI system in a professional capacity) of high-risk AI-systems have some obligations as well.

During the legislative procedure we had several public debates on the issue presenting a protestant view on AI and exchanges with members of the European Group on ethics, the German Ethics Council (“Deutscher Ethikrat”), Commission officials, Members of the European Parliament and ecumenical partners.

We welcomed a European legal framework on AI to address significant risks for users and society and the fact that the AI Regulation introduces a classification for AI systems with different requirements and obligations tailored on a risk-based approach in form of a pyramid. That means the higher the risk, the stricter the rules.

AI applications with a high-risk potential are subject to special rules including the obligation for developers to set up risk management systems, ensure technical documentation or allow deployers to implement human oversight. AI systems with limited risk are only subject to simple transparency requirements. Certain applications posing an unacceptable risk are prohibited. These include inter alia: biometric categorization systems that use characteristics such as political, religious and ideological beliefs or sexual orientation, social scoring and predictive policing. In addition, the use of AI for emotion recognition in the workplace and in educational institutions is not permitted. AI applications that exploit human vulnerabilities, such as age, disabilities and social or economic situation, are also banned. In Parliament there was quite a controversy between the different political groups on how to deal with real time remote biometric identification.

The provisions of the AI Regulation do not apply to the military and defence sector though as Member States insisted on their competence in regulating the defence area. We had argued to also address the military use of AI (e.g. use of autonomous weapons systems) as a high-risk area in a binding legal framework. We feel that this area has a considerable impact on the security and stability of political systems and raises fundamental ethical questions.

A particular challenge for the European legislator was to keep pace with the highly dynamic development of AI technologies. There will be special regulations for generative AI, namely so-called AI models with a general purpose, including those that generate content such as texts and images. Particularly powerful AI models with systemic risks will be subject to stricter requirements. Codes of practice will be developed together with model providers and stakeholders until standards and thus harmonized European norms can be used at a later stage.

Another challenge was to effectively limit the risks of artificial intelligence on the one hand, but not to hinder innovation on the other. One solution is to allow for so-called “regulatory sandboxes”, which provide space for experiments with AI before they are brought to market.

For governance, an AI Office has been established within the EU Commission to monitor the effective implementation and compliance with the AI act. Furthermore, each Member State should establish or designate a national supervisory authority to ensure the application and implementation of the rules. The AI Office aims at enabling the future development, deployment and use of AI in a way that fosters societal and economic benefits and innovation, while mitigating risks. The Office will play a key role in the implementation of the AI Act, especially in relation to general-purpose AI models.   In case of non-compliance with the new rules, fines of varying scale are foreseen.

The legislative procedure has now come to an end. Following the vote in the European Parliament, on 21 May 2024 also the Council of the European Union approved the AI Act. The implementation will be structured into different timeframes and should in total be finalised by the end of 2030.

2. Conclusion

To conclude, the AI act is far from perfect as we might discuss at a later stage, but it is a start, and we think it was worth the effort. As EKD-office in Brussels, we supported the EU-Commission’s initiative to regulate AI since the beginnings. From our perspective, a European regulatory framework was necessary providing a legal definition of AI, as human rights and human dignity are at stake. We appreciate the risk-based approach chosen and the ban on applications posing an unacceptable risk. Throughout the legislative process, we pointed to ethical concerns and advocated for a regulation that would guarantee a human-centred and trustworthy AI that would protect human rights.

We underlined that the implementation of the AI regulation should be embedded in a broad and participatory debate on the opportunities and risks of AI technologies. This would include inter alia the involvement of all relevant stakeholders, from authorities and companies to consumer protection organisations, ethicists and civil society in advisory and expert boards adapting the AI act to further developments and advising the EU institutions.

In her State of the Union address in September 2023, the EU Commission President Ursula von der Leyen proposed the creation of an independent body to deal with the risks and benefits of AI for humanity.  It remains to be seen if that could also be a place to foster a broad public debate involving a multitude of stakeholders.

We deplore the fact that dual use systems and the military use of AI was not addressed in the AI act. Given the current political reality, we feel that this is a blind spot which needs to be taken up again. In our view, the use of lethal autonomous weapons needs to be banned as also recently stressed by Pope Francis during his intervention on the G7 summit in Italy.

It remains to be seen what the impact of the AI act will be in practice as some questions relating to governance, enforcement, participation, coherence and complementarity with existing law remain, just to mention a few.

Our former chairperson of the EKD Council, Prof. Wolfgang Huber, said – and I quote: “Accompanying transformation is a theological task.”

Adding to this, I want to echo the catholic theologian Anna Puzio in underlining that: “A theology of technologization has to work interdisciplinary, interreligious and international”. Therefore, I am grateful for the workshop today.

Given the immense effect AI will have on our way of living, working, on our societies and our idea of man combined with the highly dynamic development, churches and religious communities together with representatives from science and technology should continue to be present in the public debate on challenges and benefits of AI.

Thank you for your attention. I am looking forward to your questions and comments.