This article focuses on identifying and describing the cognitive-semantic mechanisms of euphemization and the frame models representing poverty. The study is based on a bilingual corpus of 800 media texts (420 in Russian and 380 in Uzbek) from central and regional media outlets, which yielded 63 euphemistic nominations. Cognitive-semantic annotation was conducted according to B. Warren's taxonomy (metaphor, metonymy, litotes, abstraction, generalization), along with frame-semantic modeling as proposed by C. Fillmore, enabling the identification of dominant frames within each subcorpus. The findings indicate that the Russian discourse is predominantly characterized by the frames of “temporary hardship” and “threshold/boundary,” which are realized through litotic and figurative forms (e.g., nebogatyy “not wealthy,” v trudnom material'nom polozhenii “in a difficult material situation,” za chertoy bednosti “below the poverty line”). Conversely, the Uzbek discourse is shaped by the frame of “institutional accounting,” represented through metonymic designations based on registries (e.g., temir daftar (iron notebook), ayollar daftari (women's notebook), mechr daftari (notebook of mercy)) and standardized terms (e.g., kam taminlangan (low-income), echtiyodzhmand (needy), kiyin moliyaviy vaziyatda (in a difficult financial situation)). It is demonstrated that both linguistic communities avoid direct lexemes for poverty such as Russ. bednyy / Uzb. kambagal (indigent), softening negativity through quantitative or situational descriptions. The study establishes that universal cognitive operations underlie common euphemization mechanisms, while their specific realizations are determined by the sociopolitical context and genre norms of national media discourses.