“Artificial intelligence is almost a discipline of the humanities”, Sebastian Thrun

I

In a landmark development concerning the eventual regulation of artificial intelligence (AI), the European Union embarked on this complex endeavor in June of this year. The European Parliament, responsible for examining, discussing, and adopting European legislation, approved an AI law that deserves various considerations.

This legal framework is set to be fully effective from 2024, making it the first of its kind to have such significance on this critical topic in the global public agenda. However, there are still several steps to be taken for this to happen. If all goes as planned and is realized, it could serve as a model legislation for other regions and countries approaching AI regulation. There has already been pressure within the United Nations (UN), with its Security Council recently convening to establish certain foundations on the topic at hand.

We have previously discussed in these pages that AI regulation is not an option but a genuine necessity in a time when the potential consequences of advanced AI models encompass legal, ethical, political, and even cultural dimensions that are far-reaching. That is why it is natural, as we have also noted and commented on before, that world leaders from various fields, such as Elon Musk and Yuval Noah Harari, have spoken out in recent months about the possible dangers of AI.

At this initial stage, it is necessary to address the intention to ban AI with higher risks, which requires starting from the ground of definitions. Europe aims to distinguish AI technologies and categorize them into four possible categories based on the risk each of them would pose. From those that can undergo reviews and audits, or certain clarifications, to those that would be strictly prohibited for exploiting the vulnerability of groups, assessing the reputation and integrity of individuals, or manipulating behavior, the issue of risk is fundamental as a starting point.

With this in mind, and based on a previous document prepared by the European Commission, the old continent presents a proposal whose central premise is the aforementioned issue of risk, for which, as mentioned earlier, there are four levels: unacceptable risk, high risk, limited risk, and minimal risk. Categorizing each AI product or service into these categories already appears, from this moment, as one of the central challenges that the eventual regulation will face. It should not be forgotten that the sophisticated corporations developing AI software and hardware are not just any kind of business consortium but real power players with lobbyists capable of influencing parliaments and government decision-making bodies worldwide. Their name alone carries significant weight in contemporary markets.

It seems that the coming months will be marked by intense debates in the European public opinion, but even more importantly, around the world, including major powers like the United States, economic giants in the Asia-Pacific region, and emerging countries like Mexico, Brazil, or India. It is crucial that reason and awareness prevail in the deliberation of a future that has already arrived.

II

In addition to the focus on risk, the European Union places emphasis on other specific aspects in the potential and upcoming regulation of AI. Firstly, it directly states that such regulation seeks two essential matters: the protection of fundamental rights and the safety of users. This duality carries an underlying objective: to establish complete trust when it comes to the development and adoption of AI.

The protection of fundamental rights is of the utmost importance when discussing AI regulation, both concerning traditional, what we could call “analog” rights, as well as those that can be identified as “digital” in the same argumentative line. This is not to mention “neuro-rights” as rights of a future that is already beginning to affect us. It’s a positive development that these are prioritized as a precondition for AI. However, it’s even more important that this is not just rhetoric but a clear reality.

The significant issue of big data and everything it represents in terms of personal data protection in the age of AI is critical to ensuring the full safeguarding of fundamental rights. Data is one of the most valuable assets we have in these times where we primarily operate in digital realms and spaces. If AI is to have any level of invasiveness into this data, it is essential to build containment barriers to prevent erosion of our prerogatives and public freedoms due to the actions of malicious vested interests and the wild forces of the digital market, both those with presumed legitimacy and those clearly lacking it.

Meanwhile, the safety of users is also emphasized, considering that AI operates on networks, hardware, and software susceptible to multiple attacks by hackers and cybercriminals who pursue their own murky interests to the detriment of the common good as a fundamental collective aspiration.

It’s no coincidence that the topic of cybersecurity has been gaining increasing prominence and unprecedented market value, becoming one of the sectors with the highest demand, growth, and transformation possibilities in the face of the emergence of new information and communication technologies in general and AI in particular. This is why its significance for Europe is not surprising.

Furthermore, trust in terms of the development and adoption of AI is associated with the idea of certainty that must exist when dealing with such advanced technological developments as those in question. And that certainty, at the end of the day, can only be provided by Law, starting with Constitutions and international treaties. Only a robust legal system can provide the confidence needed when interacting with such innovative but potentially dangerous tools as AI. It’s important to remember that in addition to providing order and social control, the law provides certainty to all practices carried out in a given society, at least in an ideal sense.

III

Another aspect that Europe emphasizes when designing regulation on AI is that of territoriality. There is no doubt that the application of laws across time and space remains one of the unfinished and unfulfilled challenges when it comes to information and communication technologies, especially in the context of cyberspace, modern virtual arenas, social networks, and other areas. If this is extended to the realm of AI, the level of complexity definitely increases because this type of innovation involves emerging knowledge that can be equally applicable in different spaces and levels.

In this regard, the legal framework envisioned for the coming months is intended for both public and private actors, both within the European Union and beyond if the AI system in question operates in the European market or its use impacts the sphere of someone located in Europe. If in the “analog” dimension of collective life, so to speak, there are blurry lines between the public, private, and intimate spheres, differentiation between these areas becomes even more complicated when it comes to digital life and digital rights. Therefore, great care must be taken when establishing categories that are not always as clear-cut as one might think at first.

What currently happens with a topic that can serve as a parallel, such as cybercrimes? There are several challenges due to the very nature of the offenses that occur. Consider an issue like electronic fraud, identity theft, or illegal access to networks and devices, where the unlawful act can begin in Shanghai, but the hackers, cybercriminals, and pirates may have servers in Prague, and the potential victims could be in the San Francisco Bay Area, in the United States, home to several major technology companies today. In this example, three continents would be involved, although there have been cases where all five continents are somehow implicated.

Territorial jurisdiction conflicts would quickly become evident due to the lack of an international convention and multilateral cooperation treaties in this matter. This can lead to a situation where one country criminalizes certain behaviors while another does not, hindering the investigation, prosecution, and punishment of these actions. This is not a minor issue.

This serves to illustrate that when it comes to AI, the problem arises automatically. That’s why Europe reached a consensus that when AI affects providers or users of high-risk systems, the regulatory system must be applied. When put into practice, it will be interesting to observe the dynamics by which potentially criminal behavior by agents outside the European Union is investigated.

Non-professional private uses are excluded from the regulation, although this may raise questions about the potentially harmful modus operandi of individuals or entities that may fall into this category. If we speculate, given that AI is a double-edged sword, there may be justified concerns about this exclusion.

IV

Europa’s aim in regulating AI is to maximize the principle of legal certainty, which is of utmost importance when discussing information and communication technologies in general, considering their widespread use and penetration into countless households, businesses, governments, and both public and private offices on a global scale.

While it is true that companies and economic actors may have the most significant interest in effectively implementing this, it’s also necessary to emphasize that legal certainty as a principle benefits society as a whole, including the millions of domestic AI users.

Indeed, certainty is key to instilling confidence in the highly sophisticated advancements that have occurred in AI in recent years, which will undoubtedly continue to multiply rapidly. Without certainty and trust, there can be no reciprocity when government agents impose certain duties or obligations in this field because leading by example is essential.

Furthermore, compliance with regulations is largely shaped in the realm of public policies that each nation will have to develop internally. Stating these principles is not enough; they must be accompanied by political will and a firm commitment to ensure that ethical codes are respected to the fullest extent and not reduced to mere decorative elements.

Both the legal and ethical aspects must be treated with equal objectivity if positive results are to be achieved. The proper application of legal frameworks and a conscientious consideration of both rights and obligations in an ethical context are crucial to ensuring that developments remain compliant with the rule of law, which is more necessary than ever in this ambiguous era of AI.

In this context, at the community level, the European Committee on Artificial Intelligence will emerge as a body responsible for overseeing and facilitating the implementation of the legal framework concerning AI. It aims for fluidity, effectiveness, and harmonization. Within its scope of authority, it can recommend and assess high-risk systems, becoming a center of expertise that EU members can consult when necessary.

If we aspire to establish a digital rule of law, it is crucial that legal certainty, compliance with regulations, and the proper anchoring of institutional guidelines are carried out authentically, not just on paper. AI requires clear, precise, transparent, understandable, and minimally interpretable regulation; anything less would lead to ambiguity, vagueness, and suspicions that benefit no one. This digital rule of law must always be robust and resilient against the threats posed by wild powers from the digital realm while not neglecting the analog world in which we continue to operate. It’s a monumental challenge, but one that must be confronted with determination.

V

Institutional financing to address the regulation of AI is another key aspect that cannot be overlooked in the face of the new horizons ahead. Without sufficient resources, any effort may simply remain on paper. This is why there is a provision to invest one billion euros annually in AI, which represents a good starting point given the complexity and innovation that this type of technology brings.

Another significant issue is that of machines themselves, which, thanks to science fiction, Hollywood, and dystopian and futuristic TV series, are a source of fear for many people, as they may eventually take over the world. Beyond these speculative concerns, the fact is that controlling machines not only through buttons but also through laws becomes imperative. User and consumer safety, as well as the promotion of innovation, will be top priorities in machine regulations, considering that these machines have a wide range of applications, including professional and consumer products, robots, construction machines, 3D printers, industrial production lines, lawnmowers, to name a few specific cases.

Another crucial aspect is the prohibition of biases based on gender or race, which highlights the dimensions of the fundamental right to non-discrimination. Equitable and equal treatment emerges as one of the fundamental principles in AI regulation, reflecting a commitment to rights and freedoms in general.

From another perspective, the detailed evaluation of AI systems before they enter the market is essential. This includes elements such as human oversight, transparency, robustness, accuracy, traceability, documentation, and data quality. The application of quality systems and risk management is not optional but mandatory.

Biometrics is also a topic that appears in European AI projects, and this is understandable given the advanced technology it represents, which can also be vulnerable to cybercrime. Real-time biometric identification, for example, is one of the aspects that draw attention in order to protect dignity, privacy, and personal data.

As Anders Sörman-Nilsson has stated, combining human intelligence and AI can make the world a better and more empathetic place. The steps that the European Union has taken to regulate AI comprehensively, objectively, and optimally seem to be on the right path, although it’s true that they are not perfect and may require various forms of supplementation. It will be important to see how the United Nations (UN) also takes action and plays a role in the matter, so that its efforts lead to a comprehensive international convention characterized by consensus, rationality, and a forward-looking approach. Of course, each national state must assume firm commitments from both a political and budgetary perspective, and that’s where we can assess progress in the short and medium term in relation to a topic that has already reached us.