Introduction
From February 6th to February 11th 2025, Paris hosted the most recent Artificial Intelligence (AI) Action Summit. More than 100 countries and key stakeholders participated in the summit, making it one of the most inclusive events concerned with AI. The summit was meant to enhance collaboration in the sphere of AI and discuss the rising challenges and pathways towards development.
The discussions tackled the topics of sustainability, the effect of AI on the job market, and general questions of how AI can serve the public interest. During the summit, more than 100 initiatives “to foster a trustworthy AI that is accessible to all, in the public interest” from different countries and organisations were announced. The issues of sustainability and transparency were claimed to be the central focus of the event. However, not all participants seemed to be on board with such safety-oriented framing of AI development.
-
Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet
The key takeaway from the summit was the Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet. 58 countries signed it, agreeing to work collaboratively, “ensuring AI is open, inclusive, transparent, ethical, safe, secure and trustworthy”. The document also explicitly mentioned the importance of compliance with international humanitarian and human rights law when developing new AI technologies. This is a good sign for global human rights, as acknowledgment of a threat to human rights is an important first step in the process of creating fair and just AI regulation.
The participants came up with several AI development priorities that they established and pledged to support. In short, this included prioritising accessibility, sustainability and safety of AI. As part of the above-mentioned efforts, the participants agreed to create a Public Interest AI Platform to reduce the fragmentation between current Public Interest AI initiatives and to foster collaboration between the countries and the stakeholders. Other important topics mentioned as high priority were the impact on the environment and energy and the growing influence of AI in the job market. The act also explicitly stated that the countries should aim to pay attention to the threats AI can pose to information integrity.
The United States and the United Kingdom, the first and third, respectively, in the list of countries with the most AI investment, refused to sign the document. The US did not provide a specific reason for their refusal, but Vice President JD Vance gave a speech outlining that the European approach to AI regulation did not resonate with the values of the current American administration. He cautioned Europe against an overly “safe” approach to technology development, mentioning how negatively “excessive regulation” can influence progress. The UK representative explained the refusal to sign the document due to the lack of attention to the topic of AI’s impact on national security and “global governance”. While there are discussions about whether the UK decided not to sign the document in solidarity with the US, the decision still has the potential to render a significant effect on all the European countries.
Since the development of AI has transboundary and cross-jurisdictional effects globally, it creates a complex situation in which the whole world is involved. Having two of the top three highest investing countries in AI openly refuse to acknowledge the importance of cooperation, openness, and sustainability is concerning.
Due to globalisation and open access to the internet, few can be shielded from the harmful effects of AI that was developed without any regard towards ethics and sustainability. In a world where data plays an increasingly important role in the economy and where social media can easily be used to manipulate people, it is crucial to be responsible with the technology that we create. This is why it is important to pay close attention to the way in which the global leaders regulate their AI systems and to assess what can be done to mitigate the potential harm from their actions.
-
What Threats Does AI Pose to Human Rights?
Improper use and development of AI can hinder many fundamental human rights, from the right to privacy to the right to education. Due to its diverse and broad sphere of influence, AI can infiltrate almost every aspect of our lives. However, this article will focus on the two basic human rights that can be threatened by the improper use of AI as outlined in the Universal Declaration of Human Rights: the right to receive and impart information (UDHR, Article 19) and the right to work (UDHR, Article 23).
The reason for exploring these particular human rights stems from the fact that they seem most relevant and discussed in recent years. The right to receive factual information is closely tied to internet misinformation campaigns and the claims of their influence on elections in several countries in the past decade. The right to work is an especially heated topic due to rising unemployment rates and the global financial crisis of the last few years. Moreover, both topics were mentioned in the Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet. Therefore, it is important to outline exactly how these rights can be hindered by AI and what current AI legislation can do about it.
2.1 The Right to Information
AI plays a crucial role in the spread of misinformation and disinformation online, as it can generate human-like text almost instantly, rapidly disseminating misleading narratives. Additionally, AI is capable of creating highly realistic deepfakes in the form of photos and videos, further complicating efforts to maintain information integrity and security (Shoaib, 2023). The widespread use of AI-generated disinformation, if left unregulated, has the potential to cause significant societal harm, ranging from a general erosion of trust in the media to privacy violations and personal security risks (Shoaib, 2023).
Combined with the tendency of the past decade to use “fabrics of trolls” that specialise in creating disinformation campaigns by spreading false information and pretending to be regular internet users, AI can become a powerful tool for creating false narratives. It has already proven itself useful in this sphere by being used in misinformation campaigns that concern the COVID-19 pandemic and general health topics such as vaccines (Nasiri, 2025). Moreover, what is also concerning in regard to the right to receive impartial information is that AI has the potential to do unintended harm. While disinformation campaigns can be deliberate and used for specific goals, misinformation later spread by unsuspecting users can have an even larger scope and impact.
The breach of the right to information can also lead to the breach of several other human rights, such as the right to free and fair elections (UDHR, Article 21). Nowadays, electoral campaigns take place largely online, with each short video or X thread having the potential to sway voters. Considering that the governments of authoritarian countries have been manipulating information concerning electoral campaigns, it is natural to assume that with the emergence of AI, they will start to implement the new instruments on a large scale.
2.2 The Right to Work
The right to work is being discussed more and more as the concerns over the effects of AI on the job market rise. In particular, during the Paris Action Summit, the importance of action in this sphere was solidified in the signed Statement. The major problem with using AI in the job market comes at the level of recruitment. The usage of Automated Hiring Systems opens an opportunity for bias, either purposefully created or resulting from the data on which the AI models were trained.
Due to the natural lack of algorithmic transparency, it is often impossible to look into the hiring process and determine why the model favours certain candidates over others (Rodrigues, 2020). This raises another important question about discrimination and bias. AI models are trained on the information produced by people and, therefore, can inherit the same biases expressed by people, whether intentionally or not. The same issues can threaten the right to work when the employer decides to use AI to assess the quality of work of its employees.
Another point that concerns AI in the workplace is its potential to replace certain workers, which would raise additional concerns about the potential breach of human rights. For example, in the art and design sector, workers are now being replaced with generative AI models that were taught to generate artwork based on the works of actual professionals, constituting a breach of the right to intellectual property (UDHR, Article 27). Another example is the usage of AI to replace some of the workers in key areas such as healthcare. The possibility of getting a wrong diagnosis as a result of an AI model processing your request constitutes a breach of the right to health and well-being (UDHR, Article 25).
-
Current AI regulation in Europe
- 1 European Union Artificial Intelligence Act
All the issues mentioned in the previous section are acknowledged by the European Union (EU) and by the national governments of the European states. This is why, in 2024, the first EU-level AI legislation was introduced — the European Union Artificial Intelligence Act (EU AI Act).
The EU AI Act is the world’s first regulation concerned specifically with artificial intelligence systems, giving them a clear definition and outlining the rules for their use. The document proposes a framework for the classification of AI systems based on the “risk” that they pose: unacceptable risk, high risk, limited risk and minimal risk. AI systems that pose unacceptable risk, such as social scoring systems or remote biometric recognition algorithms, are prohibited from use in the EU.
Limited risk AI is subject to light transparency obligations, while minimal risk AI is unregulated. The majority of the provisions contained within the act, however, are dedicated to outlining the regulations for AI systems classified as high risk. Those responsible for ensuring compliance with the act are the providers and developers of high-risk AI systems, including third-country providers. To assess the possible risks, they are required to disclose the information on findings of a fundamental rights impact assessment (FRIA). The act also prescribes some responsibility to the deployers of AI systems, which refers to those who produce some kind of output with the use of AI but are not its end-users.
The act also states that providers of high-risk AI systems will have to disclose the specific characteristics and functions of AI systems and outline the data that they use. This has the possibility to protect human rights in the workplace by limiting the use of AI to assess candidates or the performance level of employees.
However, many other AI systems that can be harmful will not likely be classified as high-risk and will remain unregulated. Those include, for example, systems that operate chatbots and deepfakes. Even though the deployers of such systems are obliged to inform the users that they are interacting with AI-generated content, it does not ensure information safety. Those AI systems can still generate vast amounts of false information and distribute it online, facilitating mass disinformation and misinformation campaigns.
Due to its many flaws, the act received criticism from human rights organisations that expressed their disappointment with how it addresses human rights. In particular, they contend that the regulations on the collection of surveillance data are weak, and there is a lack of regulation of the use of AI in law enforcement and migration regulation. The definition of “high risk” AI systems also seems to raise many questions. It is unclear what specific attributes make an AI system high risk and what differentiates them from the ones whose risk is unacceptable. Due to this unclear definition, industry actors can find loopholes in the legislation and use them to their advantage.
Deployers of AI systems, unlike their developers, are not obliged to assess the systems that they use in accordance with FRIA. Moreover, even for AI systems developers, the document denotes that the results of fundamental rights impact assessments should merely be listed, and no efforts to prevent the negative impact are prescribed. Additionally, the assessment process does not allow for the inclusion of stakeholders such as civil society organisations. This does not comply with the claims of transparency that are being constantly made in official AI regulation discussions and global AI events, such as the recent AI Action Summit.
Another issue is the absence of a definition of an “affected person” and, therefore, no defined rights for such a person. The phrase is used in the document several times to refer to a person who was harmed due to the usage of AI systems. However, it is unclear who qualifies for this definition and how to recognize if the person was affected. Therefore, it will make it hard to use the act in legal practice as it will be impossible to definitely prove that someone can be considered an “affected person”. According to the act, the actions individuals can take to defend their rights are limited to lodging a complaint. However, even this opportunity is described vaguely and, therefore, provides weak protection to those who might suffer from the effects of high-risk AI systems.
Not only human rights activists and organisations critique the act — researchers, in particular, in the field of risk-assessment, point out the document’s tendency to oversimplify the matters of risk, trustworthiness, and human rights (Kusche, 2024; Laux, Wachter, Mittelstadt, 2024). Some also draw attention to the instrumental use of the concept of non-discrimination, which is reinforced by the lack of access of the stakeholders to the risk evaluation process (Mazur, 2024) and undefined oversight procedures for developers of AI systems (Laux, 2024).
Moreover, Kusche argues that the act does not portray human rights as something that can be harmed by AI. It is rather used as a guide along which the ethical aspect of AI development should be assessed (Kusche, 2024). All this criticism points towards the notion that the document uses some of the important “buzzwords” without an intention to provide meaningful and exhaustive protection to European citizens.
Finally, all the regulations for high-risk AI discussed above do not apply to companies that develop AI systems for national security. That means that there are no checks, no human rights violation evaluations and no way to determine what systems are being developed and how far they go. Moreover, EU-based companies can still develop technologies prohibited for use in the EU and export them to other countries. These two last points make it obvious that the EU’s current stance on AI regulation needs serious refinement. If Europe allows its companies to export technology deemed unsafe for internal use, it cannot claim to contribute to the global defence and maintenance of human rights.
3.2 Nation Level Legislation
Currently, few national laws in European countries explicitly target AI systems. However, many countries are moving in that direction. For example, Italy’s Data Protection Authority (Garante) has taken an aggressive stance on AI-related data protection issues. In 2023, the Garante temporarily banned ChatGPT due to supposed General Data Protection Regulation violations. Another example is the Netherlands, which founded a public algorithm registry where citizens can see which AI systems are used by the government. For better or for worse, the EU AI Act will inevitably serve as a guide for the EU nations’ legislation. Therefore, it is crucial to oversee the correct application of the regulations that are outlined in the act and advocate for the act’s refinement.
Conclusion
European countries aim to take the central stage as pioneers of AI regulation by introducing the EU AI Act and hosting international events such as the AI Action Summit. However, their efforts to provide the regulations sufficient for upholding human rights do not always come across as genuine or diligent enough. The refusal of the US and the UK to bound themselves by the promises of cooperation and transparency adds more colors to an already grim picture that is the state of human rights in the face of rapid AI development. Therefore, it is of utmost importance to continue to bring attention to the issue of human rights protection in the digital sphere to shape fair and just legislation.
Bibliography
Artificial Intelligence Action Summit. https://www.elysee.fr/en/sommet-pour-l-action-sur-l-ia
Bratley, M. (2024, May 25). AI Investment Race: Discover Which Countries Are Dominating the Future of Technology. IFA. https://ifamagazine.com/ai-investment-race-discover-which-countries-are-dominating-the-future-of-technology/
Browne, R. (2025, February 11). U.S. and Britain snub international AI accord in Paris. CNBC. https://www.cnbc.com/2025/02/11/us-and-britain-snub-international-ai-accord-in-paris.html
European Commission. (2024, August 1). AI Act enters into force. https://commission.europa.eu/news/ai-act-enters-force-2024-08-01_en
Euronews with AP (2024, December 20). Italy’s privacy watchdog fines OpenAI €15 million after probe into ChatGPT data collection. Euronews. https://www.euronews.com/next/2024/12/20/italys-privacy-watchdog-fines-openai-15-million-after-probe-into-chatgpt-data-collection
EU Artificial Intelligence Act. https://artificialintelligenceact.eu/
Global AI Ethics and Governance Observatory. Netherlands. https://www.unesco.org/ethics-ai/en/netherlands
Kleinma, Z. and McMahon, L. (2025, February 11). UK and US refuse to sign international AI declaration. BBC. https://www.bbc.com/news/articles/c8edn0n58gwo
Kusche, I. (2024). Possible harms of artificial intelligence and the EU AI act: fundamental rights and risk. Journal of Risk Research, 1-14. 10.1080/13669877.2024.2350720.
Lagercrantz, O. (1015, January 31). Europe’s AI Act Stumbles Out of the Gate. CEPA. https://cepa.org/article/europes-ai-act-stumbles-out-of-the-gate/
Laux, J. (2024). Institutionalised distrust and human oversight of artificial intelligence: towards a democratic design of AI governance under the European Union AI Act. AI & Society, 39, http://dx.doi.org/10.2139/ssrn.4377481
Laux, J., Wachter, S., Mittelstadt B., (2024). Three pathways for standardisation and ethical disclosure by default under the European union artificial intelligence act. Computer Law & Security Review 53, 1-13. https://doi.org/10.1016/j.clsr.2024.105957
Laux, J., Wachter, S., Mittelstadt B., (2024). Trustworthy artificial intelligence and the European Union AI act: On the conflation of trustworthiness and acceptability of risk. Regulation and Governance, 18, 3-32. https://doi.org/10.1111/rego.12512
Mazur, J., Włoch, R. (2024). Embedding digital economy: Fictitious triple movement in the European Union’s Artificial Intelligence Act. Social & Legal Studies, 33(1), 104–123. https://doi.org/10.1177/09646639231152866
McCallum, S. (2023, April 10). ChatGPT banned in Italy over privacy concerns. BBC. https://www.bbc.com/news/technology-65139406
Milmo, D. and Courea, E. (2025, February 11). US and UK refuse to sign Paris summit declaration on ‘inclusive’ AI. The Guardian. https://www.theguardian.com/technology/2025/feb/11/us-uk-paris-ai-summit-artificial-intelligence-declaration
Nasiri, S., Hashemzadeh, A. (2025). The Evolution of Disinformation from Fake News
Propaganda to AI-driven Narratives as Deepfake. Journal of Cyberspace Studies, 229 – 250. https://doi.org/10.22059/jcss.2025.387249.1119
Rodrigues, R. (2020). Legal and human rights issues of AI: Gaps, challenges and vulnerabilities. Journal of Responsible Technology, 4, https://doi.org/10.1016/j.jrt.2020.100005
Shoaib, M. R., Wang, Z., Ahvanooey, M. T., Zhao, J. (2023). Deepfakes, Misinformation, and Disinformation in the Era of Frontier AI, Generative AI, and Large AI Models. ArXiv (Cornell University). https://doi.org/10.48550/arxiv.2311.17394.
United Nations. (1948, December 10). Universal Declaration of Human Rights. https://www.un.org/en/about-us/universal-declaration-of-human-rights
(2024, April 3). Packed with Loopholes: Why the AI Act Fails to Protect Cvivic Space and The Rule of Law. European Center for Not-for-Profit Law. https://ecnl.org/news/packed-loopholes-why-ai-act-fails-protect-civic-space-and-rule-law(2024, April 4). EU: AI Act fails to set gold standard for human rights. Article 19. https://www.article19.org/resources/eu-ai-act-fails-to-set-gold-standard-for-human-rights/
