Download Report in the links below.

The III Seminar “Emerging Challenges of Artificial Intelligence: Regulation and Human Rights,” organized by IP.rec, consolidated itself as a space for critical reflection in contrast to the dominant narrative of uncritical enthusiasm surrounding artificial intelligence (AI). The central objective of the event was to reposition AI within the realm of social, economic, and political materiality, treating it as infrastructure composed of energy, labor, resource extraction, data, opacity, and power relations. From this perspective, AI was discussed as a matter of human rights and democratic regulation, rather than as a neutral or inevitable phenomenon.

In the opening session, Raquel Saraiva, Manoela Vasconcelos (representing Cesar School), and André Fernandes situated the seminar within IP.rec’s institutional trajectory, which marks eight years of activity in Recife, articulating research and advocacy in the areas of AI, privacy, surveillance, gender and technology, child and adolescent protection, and digital platform regulation. The importance of confronting the depoliticization of technologies was emphasized, recalling that the logic of information already structures Brazilian law in instruments such as the Consumer Protection Code, the Statute of the Child and Adolescent, and the Federal Constitution, and should likewise guide the debate on AI’s legal framework. The reading of the poem “Nós latino-americanos,” by Ferreira Gullar, functioned as a symbolic framing device, associating the expansion of AI with the historical possibility of reproducing violence, exploitation, and inequality in the Global South.

The first keynote, delivered by Paz Peña, presented AI as part of an imperial energy project, linking contemporary technology and coloniality. The presentation emphasized the accelerated growth of energy consumption by data centers, the persistent dependence on fossil fuels, and the risk of a merely apparent energy transition in which renewable sources are added to fossil fuels rather than replacing them. Latin America was positioned as a territory of socio-environmental cost externalization, supplying energy, raw materials, and precarious labor to sustain the expansion of global technocapitalism.

Panel 1, dedicated to the debate on data centers and sustainability, problematized the notion of “green data centers,” considered unsustainable within the current development model. Participants highlighted that data centers are physical infrastructures that are intensive in energy and water consumption, and that discourses of efficiency and renewable use often operate as greenwashing. The National Policy for Attracting Data Centers, known as ReData, was widely criticized for prioritizing the attraction of foreign investment through tax exemptions, without adequately addressing socio-environmental impacts, regional inequalities, and regulatory gaps in environmental licensing. The Northeast region was identified as at risk of consolidating a “sacrifice zone” logic, sustained by the narrative of excess renewable energy.

In Panel 2, on epistemicide in AI, the focus shifted from isolated technical problems to the epistemic hegemony embedded in algorithmic systems. Studies were presented demonstrating how language models can reproduce racial and cultural discrimination even without the explicit use of sensitive categories, through indirect inferences and linguistic markers. The debate highlighted the erasure of local cultures, accents, peripheral knowledge, and queer languages, often treated as error or noise. It was argued that confronting digital epistemicide requires community participation, culturally situated audits, and the construction of local technological and epistemic infrastructure, with special attention to the vulnerabilities of children and adolescents in digital environments.

The second keynote, delivered by Rafael Grohmann, addressed the relationship between AI and human labor, placing data labor—such as annotation, labeling, and model training—at the center of the discussion, noting that it is predominantly carried out in the Global South under precarious conditions. It was criticized that AI regulatory proposals in Brazil largely ignore the world of work, limiting themselves to professional qualification discourses that reinforce the country’s dependent position in global value chains. Examples such as the Hollywood writers’ strike and mobilizations of cultural workers in Brazil were presented as evidence that limits on AI use result from political disputes and collective organization.

In summary, the seminar constructed a convergent diagnosis that AI cannot be understood merely as software or abstract innovation. It is a material arrangement involving energy, water, territories, precarious labor, regimes of visibility, and knowledge production. AI regulation therefore demands transparency, accountability, binding socio-environmental criteria, social participation from the design phase of technologies, and the centrality of human rights, including labor rights and children’s rights.

Democratizing AI implies contesting its material and symbolic infrastructure, preventing it from consolidating as yet another vector for deepening historical inequalities and contemporary forms of coloniality.

André Fernandes