Frontier models were constructed within a few specific cultural and linguistic contexts and validated against a particular set of values. We argue this limitation operates at three levels: technical (safety evaluations fail silently in non-English languages), theoretical (alignment frameworks inherit an implicit universalism about human values that obscures cross-cultural variation), and governance (the power to declare a model safe remains concentrated in private laboratories unaccountable to the populations they affect). Brazilian Portuguese, spoken by over 215 million people, sits at the intersection of these three failures. We propose a Brazilian research agenda structured around three fronts: (i) culturally-grounded dangerous capability benchmarks constructed participatorily in Brazilian Portuguese; (ii) adaptive evaluation methodologies that keep evaluation responsive to evolving models and locally defined risks; and (iii) a distributed safety evaluation infrastructure in which nationally accountable institutions assess frontier models before large-scale deployment.

In this paper, we introduce a new set of lexicons for expressing subjectivity in text documents written in Brazilian Portuguese. Besides the non-English idiom, in contrast to other subjectivity lexicons available, these lexicons represent different subjectivity dimensions (other than sentiment) and are more compact in number of terms. This last feature was designed intentionally to leverage the power of word embedding techniques, i.e., with the words mapped to an embedding space and the appropriate distance measures, we can easily capture semantically related words to the ones in the lexicons. Thus, we do not need to build comprehensive vocabularies and can focus on the most representative words for each lexicon dimension. We showcase the use of these lexicons in three highly non-trivial tasks: (1) Automated Essay Scoring in the Presence of Biased Ratings, (2) Subjectivity Bias in Brazilian Presidential Elections and (3) Fake News Classification Based on Text Subjectivity. All these tasks involve text documents written in Portuguese.