For an ethical and efficient EU policy and international framework on Artificial Intelligence

Resolution submitted by: JEF Political Commission 2 – Internal European Policy. Adopted by the European Congress in Paris on 26 October 2019. Re-adopted by the Federal Committee in Luxembourg on 10 April 2022. Re-adopted by the Federal Committee in Budapest, 10 November 2024.

Artificial intelligence (hereafter AI) can be defined as any machine or algorithm that is capable of observing its environment, learning, and based on the knowledge and experience gained, taking intelligent actions or proposing decisions. The applications of AI are manifold, with a significant potential to improve people’s lives and generate economic benefits for society. There are however, also risks that could emerge from these new technologies. In April 2021, the European Commission put forward a comprehensive legal framework for AI, known as the AI Act. The aim of this resolution is to take stock of this
evolution and continue to encourage European institutions together with the Member States to ensure that the development of AI upholds EU values, no matter whether it is driven by academic or industrial research or pursued by the public or private sector actors.

JEF Europe acknowledges the significant progress made by EU institutions in AI regulation, notably through the GDPR, the AI Act and the High-Level Expert Group on AI. However, gaps remain, including the need for enhanced accountability and measures to address biases, disinformation, and ethical risks. Concerns extend to the implications for democracy, education, and the economy, underscoring the importance of continuous monitoring and adaptable policies to safeguard societal interests.

JEF Europe,

  • Noting the lack of awareness in connection with the definition of AI, as not many people know how the types of AI are distributed into three groups: ANI (Artificial Narrow Intelligence), AGI (Artificial General Intelligence) and ASI (Artificial Super Intelligence). The difference between a LLMs (large language model) and generative AI is also not widely known. This information is particularly important, keeping in sight that it is crucial to fully understand where AI is heading, what are the threats they might impose and their whole nature;
  • Recalling the need for global cooperation on Artificial Intelligence (AI), by recognising its inherent transboundary, global dimension, and highlighting the conclusions of the United Nations report on Governing AI for Humanity, and the EU AI act, which is the first comprehensive regulatory framework approved for AI in the world;
    Aware that the European Union (EU) is lagging behind multiple actors across the world in terms of implementation, innovation and investment in artificial intelligence, and underlining that there is still more to be done in also a need of developing and updating adequate, practical regulations, needed to protect the fundamental rights and freedoms of every person, especially as the social implications of the massive boom in AI applications are now seen day by day;
  • Believing in the potential of these new technologies to transform lives and work practices in Europe, by advancing economic growth and improving efficiency in sectors such as industry, commerce, transport, farming, healthcare, medical care, and education;
  • Highly aware that the recent breakthroughs, especially in parallel computation and neural networks easily available to public, most notably as generative AI (GenAI), has led to changes in many aspects of life, and the only way to influence its effect on Europeans is through concerted action by the EU and by global cooperation;
  • Believing that the EU has the opportunity to exert leadership in the digital sphere by regulating AI in a human-centric way based on ethics and fundamental rights, distinct both from laissez-faire as well as authoritarian models;
  • Aware of the impact of AI on all parts of society and all sectors of the economy that carries huge promises to help solve societal challenges but also has already come comes with massive instances of disruption that we should address and mitigate, such as in education, information work, fields of arts from music to publishing, intellectual property rights, exponential increase in deepfakes and other forms of disinformation, especially in social media and politics;
  • Aware that in terms of GDP growth, estimates show that deploying the necessary key technology across Europe could create trillions in additional economic activity by the year 2030;
  • Concerned about potential biases which may, and have already, arisen in the design and application of algorithms as well as language models, specifically in the terms of gender and race for instance, by learning from and reproducing existing unequal structures;
  • Concerned about the lack of accountability and transparency behind Machine Learning (ML) algorithms, whose application in some cases has already resulted in biases regarding race or gender for instance, by learning from,reproducing and compounding existing unequal structures;
  • Encouraging support from the EU for initiatives, organizations, start-ups, and companies that are aiming to tackle the technical problems regarding artificial intelligence, due to the lack of acknowledgement of this particular branch of issues. The development of AIs poses substantial risks, mainly concerning threats from the creation of AGI (artificial general intelligence), as a consequence of their current uncontrollable nature and possible misalignment problems. There are communities and networks that are emphasizing the importance of resolving the technical issues of this topic.
  • The aforementioned technical issues include the prominent alignment problem, which is in regard to humans not being able to perfectly clarify their needs for AIs due to the vast number of factors in every topic of real life (this issue has been shown even in online games). – This problem is highly neglected and needs funding and support from the EU.
  • Keeping in mind the importance of the technological side of AI. Achieving a transparent AI, by monitoring its system or overlooking its work is nearly, if not completely impossible, due the size and complicatedness AIs (they consists and produce trillions (millions of million) of matrixes, numbers, biases and complex calculations, therefore, overlooking, understanding and intervening with this work is impossible and would take tens/hundreds of years for one AI;
  • The only way to mitigate with an AI’s biases is through influencing its training datasets. A public, large language model requires extreme sizes of data to train itself to be able to produce viable results. If the EU could provide datasets consisting of millions of sentences, words etc. regarding a certain topic (for instance: race and gender equality), the given AI could be trained on these data bases to have unbiased opinion(instead of the biased environment of Google). The transparency of these datasets is manageable through monitoring and public involvement, through amendment submitions;
  • An independent, diverse and democratically controlled group should build these datasets and every publicly accessible AI should be trained on the aforementioned data bases;
  • Acknowledging the advances made by European institutions in protecting data since the adoption of the General Data Protection Regulation (GDPR)  as well as the advances made in the policy field when adopting the AI Act in June 2024;
  • Alarmed by the escalated adoption of AI-enhanced and autonomous weapons systems (AWS) by major military powers and the resulting reduced human oversight, increased lethality, and speed and a massive increase of destructive potential of armed forces;
  • Concerned that the convergence of AI, big data, and weaponization, and mass data-driven targeting methods have enabled a new level of precision and efficiency in identifying individuals and groups marked as potential hostile targets; with potential for misuse by authoritarian regimes to suppress dissent, targeting civilians, especially when definitions of “radicalization” are vague or politically motivated;
  • Noting the fast pace of AI-enhanced biotechnology, which has already resulted in massive breakthroughs and potential in medicine, such as predicting and creating new proteins, with the downside of making bioweapons, such as synthetic viruses, significantly easier and cheaper for any actor, especially with no efficient control framework in place;
  • Noting however that the existing European legislation and policies are either insufficient or just not applicable to Robotic Process Automation (RPA) and AI;
  • Concerned about the existing legal vacuum in liability and responsibility for malfunction, clear ethical framework, harmful action, omission or other damages caused by AI in the European legal framework;
  • Acknowledging the necessity of international cooperation to address the dual-use nature of AI and ensure that its applications from military and biotechnology remain aligned with ethical standards, preventing critically dangerous misuse in terrorism and wafare, and prioiritizing transparency
    and human oversight, Approving the European Commission’s general approach in the AI Act of assessing and regulating AI based on risk;
  • Concerned that the challenges raised by AI impact numerous citizen’s rights, such as, but not limited to, privacy, safety, freedom of choice, and non-discrimination;
  • Concerned about the lack of clarity and the fragmentation of laws across the Member States that would jeopardize the development of AI goods and
    services across the Single Market;
  • Welcoming the European Commission’s intention to support Small and Medium Enterprises (SMEs) in the field of AI and robotics both structurally and
    financially;
  • Welcoming the implementation of a coordinated plan for a sustainable and ethical AI developed by the European Commission to better coordinate the development of AI between the EU and the Member States;
  • Acknowledging the need for public policy to ensure the benefits of AI reach as many people as possible, in fields such as, but not limited to, education and inclusion;
  • Concerned that AI regulations proposed in the AI Act mainly focus on “high-risk AI”, which is limited to certain technologies, not including social media and other systems, which might also present important risks. Further highlights the need for clarifications for the public (for instance, through education) in regard to the definition of AI;
  • Acknowledging the risks associated with the potential misuse and weaponization of AI, which could open doors for malicious disruptions in several key fields, including, but not limited to, state security, democratic processes, and energy;
  • Welcoming the proposal in the AI Act for the creation of European Digital Innovation Hubs (EDIHs) in every European region, to share best practices, offer support to Small and Medium Enterprises (SMEs) and public sector organizations in their uptake of AI, and give access to the European AI- on-demand platform which will contain useful tools and resources;
  • Approving the funding and scaling of innovative ideas and solutions for AI proposed in the AI Act through the AI/Blockchain Investment pilot and support programme, through calls for funding to the European Innovation Council (EIC), as well as through the creation of clusters of startups in the Horizon Europe, Startup Europe and Innovation Radar vehicles;
  • Welcoming the proposal by the European Commission for a Data Governance Act, currently being discussed in the Council of the European Union, and
    which would enable public and private actors to share open data in a manner compatible with the GDPR, to foster research and development in AI
    in Europe;
  • Welcoming the provisions in the AI Act relating to specialized training and skills development, but worries that this might not be sufficient for raising awareness of AI among the general public without the support from member states;
  • Concerned about the lack of proper support of all European languages in various AI solutions, reducing the availability, usage opportunities and potential gains of those AI solutions for European companies, citizens and organizations;

JEF Europe therefore,

  1. Calls of guaranteeing an ethical approach in dealing with Artificial Intelligence, ensuring transparency of AI systems, protection of human dignity, responsibility, impartiality and non-discrimination, reliability and liability, security and privacy, and human accountability; encourages the promotion of these principles in trade agreements and in cooperation frameworks with third parties;
  2. Calls for the implementation Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (AI Act) and ensure that other initiatives in this areas, including AI Innovation Package and the Coordinated Plan on AI, are implemented, safeguarding the safety and fundamental rights when it comes to AI;
  3. Underlines the need to prepare adequate sectoral initiatives, given their exposure to AI, including, for example, the impact of AI on democracy, education, health, wellbeing, nature stewardship, civil and political rights, judiciary, anticorruption, rule of law and law enforcement, small and medium enterprises, digital single market, taxation, civil protection, culture, and other fields;
  4. Calls on EU institutions and Member States to enact liability regulation, ensuring accountability while fostering innovation;
  5. Encourages further investment in research and innovation centered on artificial intelligence (AI), and maintaining an active agenda related to the developments – benefits and risks – around artificial intelligence;
  6. Calls on EU institutions to support Small and Medium Enterprises (SMEs) and foster innovation through targeted funding, training programs, and incentives that promote ethical AI development, while ensuring their active participation in the European AI ecosystem;
  7. Welcomes the adoption of the AI Act’s goal to ensure that AI developed and put on the market in Europe is human-centric, sustainable, secure, inclusive and trustworthy, but calls on EU institutions to add explainability as well as transparency (of AI data, algorithms and processes) to the values it wants to promote, especially for high-risk AI systems;
  8. Calls for the establishment of a mechanism that ensures trustworthiness of machine-learning algorithms through open-source tools aiming to provide clear and understandable explanations for how outcomes of AI-driven tools are established;
  9. Calls for the EU to establish and promote global ethical frameworks and treaties to regulate the development, use, and trade of AI and machine learning technologies, ensuring that these are compliant with international humanitarian law, especially in high-risk areas such as surveillance, autonomous weapons systems, and military applications;
  10. Emphasizes the need to raise awareness of AI among the general public and to integrate AI-related topics into all levels of education and lifelong learning, aiming to foster a comprehensive understanding of AI’s implications and potential risks;
  11. Further calls on the establishment of a European ecosystem for AI involving all stakeholders to ensure Europe’s digital autonomy and to be able to compete on a global level;
  12. Further calls on EU institutions to continue cooperating with the private sector active in the development of AI to ensure that it benefits European society;
  13. Notes with satisfaction that the proposed AI Act creates a number of partnerships to move quickly on foundational and application-oriented AI research, most notably the European Partnership on AI, Data and Robotics (which brings together the AI community and builds on previous public-private partnerships), the AI lighthouse for Europe (alliance of research institutions) as well as further networks of AI excellence centers, and calls for all stakeholders to work together efficiently to realize the promise of European AI leadership from lab to market;
  14. Recalls previous proposals to co-fund testing and experimentation facilities (TEFs) to accelerate the deployment and uptake of AI, and asks for a very close monitoring of all high-risk AI solutions during the testing process, to obtain robust risk assessments ahead of the decision to bring the solution to the market;
  15. Welcomes the proposal in the AI Act for the creation of European Digital Innovation Hubs (EDIHs) in every European region, to share best practices, offer support to Small and Medium Enterprises (SMEs) and public sector organizations in their uptake of AI, and give access to the European AI-on-demand platform which will contain useful tools and resources;
  16. Approves the funding and scaling of innovative ideas and solutions for AI and approves cross-border funding in Europe;
  17. Calls for the promotion of a regulatory framework, such as fair data, enabling the use and sharing of data for non-profit and scientific purposes, thus increasing the quality and availability of non-personal data;
  18. Welcomes the proposal by the European Commission for a Data Governance Act, currently being discussed in the Council of the European Union, and which would enable public and private actors to share open data in a manner compatible with the GDPR, to foster research and development in AI in Europe;
  19. Calls on the establishment of complementary initiatives – additionally to regulatory measures on AI and ethics – such as certification and codes of conduct in any AI-powered systems, with benefits for the companies acting, on a voluntary basis, in compliance with them;
  20. Welcomes the provisions in the AI Act relating to specialized training and skills development, but worries that this might not be sufficient for raising awareness of AI among the general public without the support from member states;
  21. Demands further investment and measures to mitigate the effects of automation in affected sectors e.g., in retraining of people that see their field of work displaced by rapid technological advancement;
  22. Calls on EU institutions and member states to mainstream sustainability in the development of AI technology to ensure that it contributes to net-zero sustainability goals, critically noting the increasing energy demand of data centres;
  23. Urges the EU to work in parallel on the establishment of a global ethical framework regarding AI/ML training data and processes as well as their final usage;
  24. Calls on legislators to require algorithms to actively balance out lack of data and discriminatory biases in decision-making, such as but not limited  to gender and race;
  25. Calls on the EU and its member states to establish mechanisms for citizens to quickly, simply, and without interference, being it human or automated, to file a complaint about potential breaches of the rules and principles laid out in the proposed AI Act, with the aim to provide a quick response;
  26. Underline the initiatives for restricting the use of remote biometric identification systems (commonly referred to as facial recognition technologies) to be used only under the strictest conditions when no other methods are available, and demands that other AI technologies that can be used for surveillance should be treated under equally strict conditions;
  27. Calls for the EU to work towards a Global Ban on the development, sale, and use of AI-driven mass data surveillance tools based on facial recognition and internet data intended for identifying or labeling civilian individuals or groups in military and law enforcement use to safeguard civil liberties and prevent misuse by authoritarian regimes;
  28. Demands for a Moratorium on the deployment and proliferation of autonomous weapon systems (AWS) until an enforceable, international legal framework with clear accountability mechanisms is in place, requiring rigorous human oversight on any lethal decisions made by AWS;
  29. Calls for the EU to initiate an International Treaty and applicaple protocols governing and controlling the use, creation and trade of AI and machine learning in technology used to develop weapons of mass destruction, including a strict regulatory framework for algorithms that can be used for mass data-based civilian targeting, AI-enhanced warfare, autonomous weapons systems, biotechnology setting standards for transparency, security, accountability, and compliance with international humanitarian law;
  30. Urges EU institutions and member states to remain actively seized on the matter, to be able to react fast to any emerging changes in the field of AI, and to not refrain from adding new technologies or solutions to the lists of unacceptable or high-risk AI if need be;
  31. Calls for the EU and its member states to foster and support the development of various AI solutions that properly works with all of the European languages, and to foster and support proper functionality of all languages in excisting AI solutions.