The Artificial Intelligence Systems One Year After the Chatgpt Release: the Strange Alliance of Europe and Brazil

AI Revolution Unleashed: From ChatGPT’s Debut to the Global Landscape of Generative AI

A year ago, it was difficult to foresee the impending global transformation that artificial intelligence (AI) would undergo. However, a paradigm shift occurred on November 30, 2022, with OpenAI’s introduction of ChatGPT[1], marking the onset of an era where AI has become an integral part of our lives.

ChatGPT, an AI program that engages users in human-like conversations and responds to a wide array of prompts[2], made a profound impact and was viewed as groundbreaking. It swiftly garnered over one million users within a week of its launch[3].

Between 2022 and the present day, numerous other tools have rushed into the generative AI landscape, including Google’s Bard[4], Microsoft’s Bing[5], and Anthropic’s Claude[6].

Nonetheless, it is worth noting that the spread of AI models is not proving uniform globally.

The different waves in the introduction of generative AI

Firstly, it must be noted that several countries simply block the generative AI developed in Western countries. This is the case in areas like China, Iran, Russia, and Africa. Currently, residents in these regions are, for example, unable to create OpenAI accounts to access AI-powered chatbots. There is not much to discuss here, given that this is a decision from the central Governments involved.

On the other hand, focusing on the nations where ChatGPT was originally launched, the rollout of the subsequent generation of generative AI within specific jurisdictions has been systematically staggered. This suggests that major AI companies, typically based in the U.S., initially introduced the systems (successors to the first version of ChatGPT) in specific regions before gradually expanding their reach to other areas.

Notably, two significant geopolitical regions and markets suffered an apparent ostracism: the European Union and Brazil.

The most evident example occurred in May 2023, when Google introduced Bard (a generative AI similar to ChatGPT) in over 180 countries globally, excluding Europe and Brazil[7]. It wasn’t until July 13, 2022, that Google announced the launch of Bard in these regions[8].

An analogous situation has arisen in the realm of social networks. Meta’s latest social platform, Threads, has encountered delays in its European launch[9].

The safeguards established by the EU for the digital world can be intimidating for AI companies

In the case of the European Union, such a tendency can be easily explained.

It is indeed in Europe where ChatGPT was first blocked, with the ban imposed by the Italian Data Protection Authority in March 2023 (which was then lifted in April 2023)[10]. Furthermore, in April 2023, it was announced that a task force[11] would be launched between European Authorities to promote cooperation and information exchange regarding the implementation of the General Data Protection Regulation (GDPR)[12] for the ChatGPT service.

The particular “attention” given to this topic likely arises from the robust digital legislation in Europe, which vigorously protects users’ fundamental rights. Europe was the first player to enact comprehensive regulations for protecting personal data (i.e., the GDPR). It has also introduced the Digital Services Act[13], which deals with content moderation on the web. In addition, the European Union is on the verge of enacting the Artificial Intelligence Act (AI Act),[14] which is expected to be the first worldwide legislation regulating artificial intelligence (currently, as we write this article, negotiations are ongoing, and there is hope that they could lead to the finalization of the text).

The threshold for an artificial intelligence system (or for other technology solutions) to be considered compliant with European digital legislation is then continually increasing. It is obvious that AI companies may exercise caution when introducing their products in such a market.

The Brazilian digital legislation sets benchmarks that are complex for AI players to reach

The Brazilian case is rather similar to the European one. The truth is that, for now, this country is maybe Europe’s closest friend for setting safeguards in the digital world. That can be observed in several legislative processes that occurred in Brazil in the last five years.

First, the Brazilian General Data Protection Law(LGPD), which entered into force in September 2020, is clearly inspired by the European GDPR. Despite some differences, similarities prevail. Both laws have an extensive scope, encompassing personal, material, and territorial dimensions. Furthermore, they are applicable to the processing of personal data by both data controllers and processors. They also offer specific safeguards for the processing of sensitive personal data and data pertaining to children. Additionally, similarities can be observed in the rights afforded to individuals and the responsibilities imposed upon controllers and processors[15].

As to regulating AI, the matter is currently at the center of public attention in Brazil. While the country is not a major producer of AI technology, the implications of AI in various areas, such as transportation and health, are crucial. Although the Brazilian Legislator has not pronounced a specific act that deals exclusively with the use of AI yet, the legislative process is ongoing: it began with the proposal for Bill n. 21/2020,[16] later converted into Bill n. 2338/2023,[17] which is currently under discussion before the Federal Senate. Such a Bill, similar to what the EU is discussing today, incorporates risk ratings for violations of rights considered fundamental, rules on governance and the processing of personal data, as well as envisions various punishments.

The risks faced by the big tech when launching products allegedly not compliant with the digital legislation are thus real also regarding the Brazilian territory. This is, for example, testified by the fact that in Brazil (the fifth-largest social media market in the world), Meta’s latest social platform, Threads, was introduced. Still, it has immediately drawn the scrutiny of the Brazilian Data Protection Authority.[18]

What do the European and Brazilian internal public opinions think about that?

As observed above, it cannot be excluded that a constant trend may develop wherein technological tools are introduced to Europe and Brazil by AI companies later than in countries with less stringent safeguards.

Undoubtedly, these circumstances are a source of concern in both regions, and internal “public opinions” have already begun to express discontent against an approach that might be perceived as overly protective.

In Europe, this has been made apparent for example in June 2023 when over 150 executives from various companies had criticized the AI Act in an open letter[19] sent to the European Parliament, Commission, and Member States, arguing that the AI Act could undermine Europe’s competitiveness and technological sovereignty. The letter asserted that companies developing AI systems might face excessive compliance costs and liability risks, potentially prompting major players in the field to completely withdraw from the European market.

In Brazil something very similar happened. In early November 2023, a coalition of entities, including trade association federations (Coalition for Innovation and Responsibility in Artificial Intelligence), issued an open letter,[20] directed at the Senate Committee overseeing the matter. The letter raises concerns about the aforementioned Bill 2338/23. According to the group, the adoption of a legal framework featuring a strict system of responsibility and sanctions, as proposed by such Bill, would position Brazil’s regulation as among the most stringent and restrictive globally. This would then raise the risk of discouraging research, development, and innovation activities, further distancing the country from a leading role in AI.

Conclusions

Considering all the above, the final question is whether the robust legislation for the protection of fundamental digital rights, adopted by both Europe and Brazil, pays off.

One may argue that the European and Brazilian Legislators should lower down the guarantees for avoiding the risks that the two countries are left “behind” (embracing what a certain part of their public opinion is currently flagging as we have seen above).

On the other hand, the counterargument is that Europe and Brazil could play a fundamental role in raising awareness of the necessity to develop AI systems (including the generative AI) that respect the fundamental digital rights.

As a matter of fact, European and Brazilian strict digital legislation have the merit of compelling the big tech companies to develop solutions respectful of the users’ digital rights before entering their markets (Europe and Brazil are indeed two markets too significant to be abandoned).

While the current delay in introducing certain technological solutions in these territories may be perceived as a disadvantage today, it has the potential to become an advantage tomorrow. The stringent parameters set by Europe and Brazil indeed ensure the consistent delivery of artificial intelligence systems in these regions that are respectful of human rights and fundamental freedoms, even in the digital environment.

[1] See Introducing ChatGTP, on openai.com, 30 November 2022.

[2] Please review OpenAI’s website.

[3] See K. Buchholz, Threads Shoots Past One Million User Mark at Lightning Speed, on statista.com, 7 July 2023. Interestingly, in such report it is indicated the time it took for selected online services, including ChatGPT, to reach one million users.

[4] The description of the tool Bard is available on Google’s website.

[5] The tool Bing is available on Microsoft’s website.

[6] The tool Claude is available on Antrophic’s website.

[7] The list of countries, updated as of May 2023 and excluding the European Union and Brazil, is available on Google’s website.

[8] Bard is currently available in more than 40 languages and over 230 countries and territories.

[9] Interestingly, in July 2023, Rob Sherman, Meta’s chief privacy officer for policy, wrote on Threads that «the app does meet GDPR requirements today. But building this offering against the backdrop of other regulatory requirements that have not yet been clarified would potentially take a lot longer, and in the face of this uncertainty, we prioritized offering this new product to as many people as possible».

[10] Italian Data Protection Authority, 30 March 2023, n. 112 (doc. web n. 9870832).

[11] European Data Protection Board, EDPB resolves dispute on transfers by Meta and creates task force on Chat GPT, on edpb.europa.eu, 13 April 2023.

[12] Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data and repealing Directive 95/46/EC (General Data Protection Regulation).

[13] Regulation (EU) 2022/2065 of the European Parliament and of the Council of 19 October 2022 on a Single Market for Digital Services and amending Directive 2000/31/EC (Digital Services Act).

[14] Proposal for a Regulation of the European Parliament and of the Council laying down harmonized rules on Artificial Intelligence (Artificial Intelligence Act) and amending certain Union Legislative Acts.

[15] Rennò Penteado Sampaio Advogados, Brazilian Data Protection Law (LGPD, English Translation) (As amended by Law No. 13,853/2019), on iapp.org, October 2020.